Substitution Reactions in the Pyrolysis of Acetone Revealed through a Modeling, Experiment, Theory Paradigm

pubs.acs.org/JACS

Article

Substitution Reactions in the Pyrolysis of Acetone Revealed through

a Modeling, Experiment, Theory Paradigm

Daniel P. Zaleski, Raghu Sivaramakrishnan, Hailey R. Weller, Nathan A. Seifert, David H. Bross,

Branko Ruscic, Kevin B. Moore, III, Sarah N. Elliott, Andreas V. Copan, Lawrence B. Harding,

Stephen J. Klippenstein, Robert W. Field, and Kirill Prozument *

Cite This: J. Am. Chem. Soc. 2021, 143, 3124 −3142

Read Online

sı

*

Metrics & More

Article Recommendations

Supporting Information

ACCESS

ABSTRACT: The development of high-ﬁdelity mechanisms for

chemically reactive systems is a challenging process that requires the

compilation of rate descriptions for a large and somewhat ill-deﬁned

set of reactions. The present uniﬁed combination of modeling,

experiment, and theory provides a paradigm for improving such

mechanism development eﬀorts. Here we combine broadband

rotational spectroscopy with detailed chemical modeling based on

rate constants obtained from automated ab initio transition state

theory-based master equation calculations and high-level thermo-

chemical parametrizations. Broadband rotational spectroscopy oﬀers

quantitative and isomer-speciﬁc detection by which branching ratios

of polar reaction products may be obtained. Using this technique, we

observe and characterize products arising from H atom substitution

reactions in the ﬂash pyrolysis of acetone (CH₃C(O)CH₃) at a nominal temperature of 1800 K. The major product observed is

ketene (CH₂CO). Minor products identiﬁed include acetaldehyde (CH₃CHO), propyne (CH₃CCH), propene (CH₂CHCH₃), and

water (HDO). Literature mechanisms for the pyrolysis of acetone do not adequately describe the minor products. The inclusion of a

variety of substitution reactions, with rate constants and thermochemistry obtained from automated ab initio kinetics predictions and

Active Thermochemical Tables analyses, demonstrates an important role for such processes. The pathway to acetaldehyde is shown

to be a direct result of substitution of acetone’s methyl group by a free H atom, while propene formation arises from OH substitution

in the enol form of acetone by a free H atom.

iterative reﬁnement of selected parameters within the models

(often by guidance from sensitivity and path ﬂux analyses²) to

reproduce the available experimental data.³These reﬁnements

commonly occur over multiple years as ever more detailed

experimental and theoretical data provide increasingly

stringent constraints on the process. Here we illustrate the

utility of a direct coupling of modeling, experiment, and theory

(MET) in a detailed analysis of the pyrolysis of acetone.

Acetone ﬁnds widespread use as a solvent and as an

intermediate in numerous chemical processes. With a near

ubiquitous presence in the troposphere as a result of both

biogenic and anthropogenic emissions,⁴its formation and

degradation kinetics is of relevance to atmospheric chemistry.

Acetone is also formed as an intermediate during the

1. INTRODUCTION

Reactive environments, such as those that occur in fuel

combustion, planetary atmospheres, chemical vapor deposi-

tion, and particle synthesis commonly involve a large number

of chemical reactions. Indeed, a quantitative description of the

conversion of a fuel into both products and pollutants generally

requires a chemical model with ∼10³species and ∼10⁴

reactions (e.g., a recent combustion mechanism for isooctane

contains 2768 species and 9220 reactions¹). Such highly

descriptive chemical models are of utility in studying the

response of the reacting systems to various perturbations. For

example, within combustion, one is often interested in the

response of pollutant concentrations and/or ignition properties

to changing engine design parameters, or to fuel additives. Of

course, such models are most useful when they accurately

reproduce the true conversion routes.

Received: November 6, 2020

Published: February 22, 2021

With so many reactions occurring simultaneously it is

generally quite challenging both to enumerate all the

signiﬁcant reactions and to obtain high quality estimates for

all the mechanistic parameters. Traditionally, modelers have

developed chemical kinetics models for such processes by

Operator of Argonne National

Laboratory. Published by American

Chemical Society

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

3124

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

combustion of many hydrocarbons. Furthermore, it is widely

used as a fuel tracer⁵to facilitate noninvasive, in situ, laser-

induced ﬂuorescence imaging of fuel molecules in combustion

systems.

chemically activated intermediate (CH₃CH(O)CH₃*) that

undergoes “direct” dissociation (without stabilization) to the

products CH₃+ CH₃CHO in ≲1 ps. In the physical chemistry

and combustion kinetics literature,⁴¹⁻⁴³such processes are

typically referred to as “well-skipping”, “formally direct”, or

“chemically activated” reactions. Replacement of a functional

group with an H atom^44,45or another radical^46,47is called an

“addition−elimination”⁴³or a “homolytic substitution”⁴⁸

reaction. In this work, we choose the term “well-skipping

substitution” to emphasize the transient nature of the

nonstabilized intermediate complex CH₃CH(O)CH₃*, and

also abbreviate the term to “substitution”. In Section 3.4.2 we

consider a hypothetical scenario in which CH₃CH(O)CH₃is

initially stabilized and show that it would still be short-lived at

high temperatures. Related well-skipping substitution reactions

of the enol provide routes to enhanced production of propene

as well as OH and then H₂O, as detailed below. These

modeling predictions were validated through subsequent direct

experimental observation of propene and water. The

theoretical studies performed here help highlight the explicit

role of “well-skipping” pathways to the observed products in

the present studies.

The implementation of these new substitution channels in

the kinetics models requires quantitative descriptions of the

corresponding rate constants and species thermochemistry.

Recent progress in theory allows for the automated prediction

of this data by ab initio kinetics.^49,50Here we applied these

automated kinetics methods to the prediction of the rate

constants for the full set of relevant well-skipping reactions, as

well as several other reactions that were of speculative

signiﬁcance. Incorporation of these ab initio kinetics

predictions within the model ensures that it is based on

physically sound kinetic parameters, thereby ensuring the

validity of the conclusions drawn from the modeling. To

further enhance the ﬁdelity of the core model, we also

performed Active Thermochemical Tables analyses^51,52for the

species of primary importance to the pyrolysis.

Pyrolysis is of both academic⁶⁻¹⁵and commercial^16,17

interest. Furthermore, mechanisms for pyrolysis are an

important subcomponent of combustion mechanisms. Pyrol-

ysis of acetone, one of the simplest ketones, has long drawn

attention.¹⁸⁻²¹At temperatures below ∼1000 K, acetone

decomposition is a prototypical example of a free-radical

mechanism involving chain reactions.²²The mechanistic

aspects outlined in the classic Rice−Herzfeld mechanism²²

nearly 100 years ago have largely stood the test of time in

describing the detailed kinetics of acetone decomposition²³

even at higher temperatures.

For the present eﬀort, the relative simplicity of the pyrolysis

of acetone provides a convenient target for a combined MET

analysis. Indeed, it is so simple and so well studied that one

might wonder whether the large body of prior work would

preclude any meaningful new observations. However, prior

studies of acetone pyrolysis at higher temperatures (>1200 K)

have largely been restricted to time-resolved absorption

measurements of select intermediates or stable species proﬁles

at long residence times (∼2 ms) in shock-tube studies.²⁴⁻²⁶As

a result, we ﬁnd that signiﬁcant reactions have been neglected

in prior work, with these reactions being the primary source for

important secondary products observed for the ﬁrst time here

in the high-temperature pyrolysis of acetone.

The experimental aspect of the present eﬀort involves direct

detection and quantiﬁcation of a number of the key acetone

pyrolysis reaction products that emerge from a silicon carbide

(SiC) microreactor.^6,7,27We use Chirped-Pulse Fourier

Transform rotational spectroscopy^28,29in the millimeter-wave

region (CP-FTmmW).^30,31Since the frequencies of rotational

transitions principally depend on the molecular carrier’s three

moments of inertia, rotational spectroscopy achieves chemical

speciﬁcity, distinguishing between isotopologues, isomers, and

conformers.³²Because CP-FTmmW spectroscopy involves

multiplexed detection and simultaneously samples ∼10⁴

spectral resolution elements, it is ideal for detection of

transitions that belong to multiple chemical species.^33,34In

addition, transitions observed using CP-FTmmW spectroscopy

maintain meaningful relative intensities.^28,35This intensity

information can then be converted into relative abundances of

observed compounds.³⁶⁻³⁸An additional feature of the SiC

reactor is the short observation times (25−150 μs), which

facilitates direct access to the nascent steps involved in

pyrolytic processes at high temperatures. These key aspects are

utilized here to study reaction kinetics and dynamics in a ﬂash

pyrolysis reactor.³⁹

2. METHODS

2.1. Experiment. All rotational spectra were recorded using the

Argonne W-Band (75−110 GHz) CP-FTmmW spectrometer built by

BrightSpec.^31,53,54One of the primary challenges with broadband

rotational spectroscopy is susceptibility to spurious signals.⁵⁵These

unwanted signals can arise from the electronics or by frequency

mixing between the electronics’ “spurs” and the molecular emission

itself. To overcome these obstacles, the segmented CP-FTmmW

spectroscopy²⁹was developed and is implemented in the BrightSpec

spectrometer as High Dynamic Range mode. In this regime, narrow-

band chirps polarize the molecular rotational transitions in small, 30

MHz, segments of the 75−110 GHz band. The fast digitizer acquires

the free induction decays (FIDs) from a 30 MHz segment and

averages them before moving on to the next segment.³¹The spurious

signals caused by the molecular emission mixing with, for example, the

local oscillator are absent from the narrow 30 MHz windows. The 25

mW excitation chirps are 0.5 μs in duration, and each FID is collected

for 4 μs. Averaged FIDs are Fourier transformed after application of a

Kaiser−Bessel digital ﬁlter to improve baseline resolution.⁵⁶Another

important advantage of the segmented chirp technique and its

implementation by BrightSpec is almost full (80%) acquisition duty

cycle.³¹The spectrometer provides an optimal temporal overlap with

steady state molecular samples, which we implement in this work

using a continuous (CW) supersonic expansion from the heated SiC

microreactor.

The present experimental observations, particularly that of

acetaldehyde, led the modeling eﬀort to explore the

importance of bimolecular homolytic substitution (S_H2)

reactions between the fuel (acetone) and H, CH₃, and OH

radicals. These substitution reactions were considered for

acetone in both its ketone and enol forms (2-propenol). This

choice was stimulated by our prior work, which indicated an

important role for the enol isomer in the pyrolysis of

acetaldehyde.⁴⁰Notably, the present study indicates that

acetaldehyde, which is observed experimentally here for the

ﬁrst time in acetone pyrolysis, arises from H atom addition to

acetone, followed by CH₃loss. Importantly, this is not a

sequential process, but rather involves the formation of a

Molecular sources for rotational spectroscopy generally involve the

use of free jets that originate from a pulsed solenoid valve.³⁹Pulsed

3125

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

introduction of gas sample to the microreactor can cause turbulent

ﬂows and could present a challenge for the coupled ﬂow/kinetic

modeling.⁵⁷⁻⁵⁹To address that limitation and to achieve the optimal

temporal overlap of the molecular sample with the nearly continuous

mode of FID acquisition in the segmented chirp spectrometer, a CW

supersonic expansion is employed in this work. A 6 cm, 1.5 mm inner

diameter SiC tube is held in place by a vespel/graphite ferrule in a 1/4

in. Swagelok ﬁtting. As in ref 60, two carbon discs in contact with

molybdenum electrodes and separated by 3 cm connect the SiC tube

to the electrical power supply. A water baﬄe provides further support

for the SiC tube and thermally isolates it from the gas supply line. The

these prior eﬀorts to guide the closed, homogeneous reactor

simulations of the present experiments.

2.3. Theory. Our recent automation of the process for generating

ﬁrst-principles kinetic predictions for large sets of reactions^49,50

allowed us to readily predict the rate coeﬃcients for essentially all of

the key reactions involved in the initial stages of pyrolysis. In

particular, the methodology proceeds directly from a listing of

reaction channels and desired kinetic methodology to Chemkin

formatted rate constants and thermochemical properties. The

methodology spawns all of the necessary electronic structure

calculations including geometry optimizations, TS and conformational

searches, as well as hindered rotor scans, vibrational analyses, and

high-level energy computations. This data is then automatically

collected and formulated for the implementation of multichannel

multiwell master equation analyses, as well as the requisite ﬁtting of

the rate constant data to forms appropriate for use in Chemkin style

modeling simulations. A brief review of this automated kinetics

methodology is provided in the Supporting Information 2.

The kinetic methodology implemented here involved transition

state theory (TST) based master equation calculations, with the

barrier properties predicted at either the CCSD(T)-F12/CBS//

B2PLYP-D3//cc-pVTZ or the CCSD(T)-F12/cc-pVTZ-F12//

ωB97XD//cc-pVTZ levels of electronic structure theory. The

complete basis set (CBS) extrapolation in the former analysis is

based on explicit evaluations for the cc-pVTZ-F12 and cc-pVQZ-F12

basis sets. The ﬁrst level of theory was applied to reactions on

potential energy surfaces with 4 or fewer heavy atoms (e.g., H + 2-

propenol and H + acetone), while the second level was applied to

potential energy surfaces with 5 heavy atoms (e.g., CH₃+ 2-propenol

and CH₃+ acetone). The results reported in ref 50 indicate that

calculations at these levels of theory typically have an uncertainty of

about a factor of 2 or better in the predicted kinetics. For a few

reactions we instead performed CCSD(T)/CBS(cc-pVTZ,cc-

pVQZ)//M06-2X based ab initio TST based master equation

calculations by hand,^63,64which are expected to be of similar accuracy

to the CCSD(T)-F12/cc-pVTZ-F12//ωB97XD//cc-pVTZ based

ones. These electronic structure calculations were performed using

the Gaussian 09 software package.⁶⁵All of the structures, vibrational

frequencies, and energies from this analysis are provided in the

Supporting Information 2.

A second aspect of the theoretical analysis involved a high-level

determination of the potential energy surface properties for the

dissociation of acetone. This analysis was mostly performed at the

ANL0 level of theory, as described in ref 66. The extensive

comparison with Active Thermochemical Tables values provided

therein indicated that ANL0 predicted stationary point energies

should generally have uncertainties of ∼0.2 kcal/mol. By necessity,

this comparison was only for minima, and it is less clear how accurate

saddle point predictions would be. Nevertheless, more limited

unpublished comparisons between related kinetic predictions and

observations suggest that the accuracy for saddle points is quite

similar, at least for “standard” saddle points with limited multi-

reference character.

For the roaming saddle points, CCSD(T) based methods are not

appropriate, and the energies were instead obtained from multi-

reference based estimates of the energy relative to the fragments. In

particular, the energies were obtained from second order multi-

reference perturbation theory (CASPT2) determinations of the saddle

point structure and frequencies (for a cc-pVQZ basis set), and

singlet−triplet splittings (at the CBS limit from cc-pVQZ and cc-

pV5Z calculations). The singlet−triplet splitting is then added to a

CCSD(T)-F12/CBS prediction of the triplet state energy at the

saddlepoint, which in turn is referenced back to the corresponding

energy of the radicals. These energies are then tied back to the

acetone energies through ANL0 values for the radicals. For

comparison purposes, Davidson corrected multireference conﬁg-

uration interaction (MRCI) values of the spin-splitting are also

obtained. A 4 electron 4 orbital active space (correlating with the

radical orbitals and the CO pi orbitals) was employed for the

reference space for all multireference calculations.

SiC microreactor is heated to its wall temperature of around T_wall

=

1800 K as measured by an optical pyrometer. A CW microreactor has

been previously implemented with other detection techniques.⁷For

rotational spectroscopy, eﬃcient adiabatic cooling is essential to

increase the S/N. In this work we seed the acetone precursor

molecules in argon carrier gas before feeding this mixture to the

heated SiC microreactor. The supersonic expansion of this mixture

from the microreactor must extend for at least several centimeters

beyond the reactor’s exit without being destroyed by the gas in the

vacuum chamber. In that case, when the millimeter-wave beam (3 cm

in diameter) crosses the molecular beam about 4 cm downstream, the

spectrometer samples the rotationally cooled pyrolysis reaction

products. This is achieved by ample pumping applied to the vacuum

chamber by two turbomolecular pumps, Osaka TGkine1704MBWC-

B and Osaka TG900MCWC,³¹with the jet pointing directly into the

latter. Mass ﬂow controllers (MKS Instruments, GM50A) maintain

stagnation pressures of p₀= 550 mbar at 1000 sccm ﬂow rate of the

Ar/acetone mixture, while the pressure inside the vacuum chamber is

p_ch= 1.3 × 10⁻²mbar. The mixtures were prepared using acetone

(≥99.9%, Sigma-Aldrich), acetone-d₆(99.9% D atom, Sigma-Aldrich),

and argon (99.9999%, Airgas). In these pyrolysis studies we record

two consecutive spectra in order to eliminate the possibility of false

detection due to impurities. The “pyrolysis” spectrum is recorded with

the microreactor heated and contains the rotational transitions of the

products and intermediates of thermal decomposition. These spectral

lines must disappear in the “background” spectrum, which is recorded

with the reactor at room temperature.

2.2. Modeling. A kinetics model consisting of 67 species and 242

reactions was assembled to describe the pyrolysis of CH₃C(O)CH₃.

The subsequent sections outline the mechanistic steps describing the

relevant unimolecular and bimolecular reactions of interest in acetone.

The kinetics of the primary unimolecular and bimolecular reactions,

essential to the early stages of high temperature acetone pyrolysis, was

revisited in this work. In particular, kinetics for the bimolecular

reactions of reactive radicals (CH₃, H, OH) with acetone and its enol

isomer, and for their immediate products are computed using an

automated theoretical kinetics code developed in-house as described

in the theory section. Secondary chemistry essential to describe the

formation of various intermediates relied on our in-house

mechanism^40,61to describe the pyrolysis and oxidation of small

hydrocarbons. The complete kinetics model, including thermochem-

istry for all the species (represented by NASA polynomials) used for

the present acetone pyrolysis simulations, is provided in the

Supporting Information.

The acetone pyrolysis simulations reported here utilized the

ANSYS Chemkin-PRO⁶²suite of programs. The predictive capability

of our acetone kinetics model at high temperatures relevant to the

present SiC reactor studies was tested against simulations of time-

resolved species proﬁles from absorption measurements behind

reﬂected shock waves by Lam et al.²⁴and speciation data²⁵from a

single-pulse shock tube. The results of these simulations are reported

in the Supporting Information (Figures S1 −S6), where the model is

found to accurately reproduce these prior observations. This

validation lends support to the mechanistic interpretations and

simulations of the species formed in the present SiC microreactor

experiments. The role of potentially turbulent ﬂow-ﬁelds in the SiC

microreactor has been highlighted in the experimental Section and in

prior eﬀorts.⁵⁷⁻⁵⁹In this work we utilize the insights gained from

3126

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

a

Table 1. Energetics (kcal/mol) for the Dissociation of Acetone

b

c

d

e

species

ATcT

ANL0

CCSD(T)//M06-2X

CCSD(T)/B3LYP

CCSD(T)//MP2

CH₃C(O)CH₃

0

CH₃CO + CH₃

CH₂C(O)CH₃+ H

CH₃CHO + CH₂

CH₃CCH₂+ OH

82.80 0.09

95.01 0.16

104.21 0.08

119.56 0.16

112.62 0.22

−

82.6

95.2

104.3

119.7

112.7

64.9

82.4

95.5

81.0

83.9

96.3

119.8

114.8

66.7

122.1

117.1

65.5

CH₂COH + CH₃

TS

65.5

13.2

CH₃C(O)CH₃↔ CH₂C(OH)CH₃

CH₂C(OH)CH₃

11.63 0.16

−

11.5

86.9

12.9

11.5

87.7

TS

CH₃C(O)CH₃↔ CH₂CO + CH₄(4-centered)

f

TS

−

81.5, 81.6

81.6, 82.3

84.8

CH₃C(O)CH₃↔ CH₂CO + CH₄(roaming)

TS

−

96.8

21.0

97.3

98.8

97.0

22.0

CH₂C(OH)CH₃↔ CH₂CO + CH₄

CH₂CO + CH₄

21.05 0.07

21.5

21.30

TS

−

117.0

CH₃C(O)CH₃↔ C₂H₆+ CO

f

TS

−

87.2, 87.8

92.7

CH₃C(O)CH₃↔ C₂H₆+ CO (roaming)

C₂H₆+ CO

4.26 0.07

4.1

6.2

TS

−

90.0

91.4

93.3

91.3

CH₂C(OH)CH₃↔ CH₃CCH + H₂O

CH₃CCH + H₂O

36.75 0.08

−

36.7

90.6

37.1

93.3

39.5

94.3

38.7

92.0

TS

CH₂C(OH)CH₃↔ CH₂CCH₂+ H₂O

CH₂CCH₂+ H₂O

37.88 0.08

37.8

38.1

40.4

39.2

TS

a

Chemical reactions connected with

↔

are associated with the saddle point energies. Other entries display energy diﬀerences between the listed

b

products and CH₃C(O)CH₃. Computational energies include zero-point corrections. Δ_rH° (0 K) for the listed decomposition reactions of

acetone from Active Thermochemical Tables, using the Thermochemical Network ver. 1.122v. CCSD(T)/CBS(cc-pVTZ,cc-pVQZ)//M06-2X/

cc-pVTZ. CCSD(T)/6-311+G(3df,2p)//B3LYP/6-311G(d,p) from ref 81. CCSD(T)/aug-cc-pVDZ+2df//MP2/6-311++G(2d,2p) from ref 82.

The primary entries are CASPT2 based entries, while the numbers in italics denote Davidson corrected MRCI based results. Both are tied back to

c

d

e

f

the acetone energy through the ANL0 value for CH₃+ CH₃CO as described in the text. Note that there are two saddle points for the roaming

process that leads to CH₂CO + CH₄. The ﬁrst of these (top row) allows for reorientation of the CH₃moiety from one side to the other, while the

other (bottom row) corresponds to the saddle point for abstraction of the H atom.

2.4. Thermochemistry. For the most important species, it was

3. RESULTS AND DISCUSSION

deemed useful to additionally corroborate the computational results

3.1. Theory of Unimolecular Reactions of Acetone.

with values obtained from Active Thermochemical Tables

Whether via pyrolysis,^{11,15,20,23,24,70−73}shock wave,^25,26,74or

(ATcT).^51,52To this end, the reaction enthalpies and free energies

photolysis,⁷⁵⁻⁸⁰decomposition of acetone has been well

for all of the primary reactive species were independently obtained

studied. It has long been understood that, below about 1100

from ATcT using the most current Thermochemical Network (TN),

K, the pyrolysis of acetone proceeds through a free methyl

ATcT TN ver. 1.122v. This TN was developed by successive

radical mechanism.^18,22Nevertheless, it is instructive to review

expansions of the earlier⁶⁷ATcT TN ver. 1.122r, in order to include

from a chemical perspective how the pyrolysis of acetone leads

all of the species relevant to several ongoing studies, including the

to its decomposition into successively smaller species. This

present one. The current ATcT TN ver. 1.122v describes the

discussion is a key prelude to understanding the variety of

thermochemistry of more than 2300 chemical species, interconnected

by ∼29 000 thermochemically relevant determinations, based on both

experiment and theory. For the primary reactive chemical species

routes that might contribute to the experimentally observed

minor products. Automated theoretical kinetics is then used to

obtain high quality rate data for the previously unexplored

mentioned in this study, the temperature-dependent thermophysical

channels, which the modeling employs in exploring the relative

properties, such as the heat capacity, the entropy, and the enthalpy

contribution of the various pathways.

incrementthe latter two important in deriving correct thermochem-

The ﬁrst step in the pyrolysis of a fuel generally involves

istry at 1800 Kwere derived from nonrigid-rotor-anharmonic-

either its thermal molecular decomposition, or, if there is a

suﬃcient density of radicals present, the abstraction of an H

atom from the fuel. For simplicity, we start by considering ﬁrst

oscillator (NRRAO) partition functions, which were constructed by

combining closed expressions for the rigid degrees of internal

motion⁶⁸with direct counts for large amplitude motions, such as

the thermal decompositionthe abstraction reactions natu-

internal rotors. The quoted uncertainties, which formally pertain to

the 298.15 K results and correspond to earnest estimates of the 95%

conﬁdence limits,⁶⁹were obtained by using the full ATcT covariance

matrix. Although the uncertainty of the thermochemistry at 1800 K is

expected to be moderately larger, it is not expected to be drastically

dissimilar to that given for 298.15 K, considering that the underlying

partition functions are of NRRAO quality.

rally arise from the products of the thermal decomposition.

The thermal decomposition itself generally involves a

competition between simple bond ﬁssion and molecular

elimination. Furthermore, isomerization to a new chemical

structure may also precede the decomposition into two

molecular fragments. For acetone, the possibly relevant

decomposition channels are

3127

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Figure 1. Rotational spectra of ketene and acetaldehyde detected as the acetone pyrolysis products. Acetaldehyde is a signature of the discovered

substitution reaction (R19). In this experiment, the 0.5% acetone in argon (by mole) are ﬂown through the T_wall= 1800 K microreactor, and, along

with the reaction products, adiabatically cooled in the supersonic expansion for millimeter-wave detection. (Top Panel) The black trace is the 75−

110 GHz spectrum after averaging 10⁶FIDs from each 30-MHz segment. The S/N of the strongest transition is 7500:1. (Middle Panel) The black

trace is the residual pyrolysis spectrum after acetone and its isotopologues have been removed by postprocessing. The blue trace is a T_rot= 15 K

simulation of ketene transitions. The red trace is a T_rot= 5 K simulation of acetaldehyde transitions. (Bottom Panel) An 800× vertical and a 500×

horizontal magniﬁcation highlights the 4₀₄−3₀₃transitions of acetaldehyde. The rotational transition is split into the E and A symmetry components

due to the interaction of the CH₃internal rotor levels in the ground torsional state. The experimental data are available in the SI.

and (R4) at suﬃciently high temperatures. The contribution

CH₃C(O)CH₃→ CH₃CO + CH₃

from (R5) is expected to be completely negligible (see also

Section 3.5), even at high temperatures, since its products lie

30 kcal/mol higher in energy than those from (R1).

(R1)

CC fission

CH₃C(O)CH₃→ CH₃C(O)CH₂+ H

Roaming radical abstractions,⁸⁴where the incipient CH₃

radical reorients (roams) while dissociating and then abstracts,

provide alternative pathways for the (R3) and (R4) molecular

eliminations. The highest energy saddle point for the CH₃

roaming to abstract an H channel is predicted to lie

1.0 kcal/mol below the asymptotic energy of CH₃CO +

CH₃. A previous study⁸²of the CH₃roaming barrier employed

single reference electronic structure methods, which are

inadequate⁸⁵for treating such long-range radical−radical

interactions. As a result, their calculations erroneously predict

a roaming barrier that is 1.0 kcal/mol above the radical

asymptote. Qualitatively, one expects the CH₃roaming process

to provide a minor production of CH₄during acetone

dissociation (perhaps 10% at most for experimentally relevant

temperatures, with that amount decreasing with increasing

temperature^85,86). Furthermore, the analogous roaming

channel in CH₃CHO decomposition was previously ob-

served⁴⁰to be of minor signiﬁcance in simulating product

distributions. Thus, for simplicity the roaming channels were

not included in the present modeling of acetone decom-

position.

(R2)

CH fission

CH₃C(O)CH₃→ CH₂CO + CH₄

CH₄elimination

(R3)

(R4)

CH₃C(O)CH₃→ C₂H₆+ CO

C₂H₆elimination

CH₃C(O)CH₃→ CH₃CHO + CH₂

CH₂elimination

(R5)

(R6)

CH₃C(O)CH₃→ CH₂C(OH)CH₃

keto−enol isomerization

The reaction energetics, as well as the transition state energies,

for these processes are summarized in Table 1.

Since the saddle points for methane, ethane, and methylene

elimination channels lie at higher energy than the CC ﬁssion

channel (Table 1) and have lower entropies, simple CC ﬁssion

(R1) is expected to dominate the kinetics. Indeed, a prior

shock tube study found good agreement with theoretical

predictions based solely on a model for the CC ﬁssion

process.⁸³Nevertheless, there may be some minor role (e.g.,

1% of the dissociative ﬂux) from the elimination channels (R3)

From Table 1, it is also clear that the keto−enol

isomerization is the energetically most facile process from

acetone, and so we should also consider its possible

dissociation fragments. However, prior studies⁴⁰of the

3128

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

decomposition of the related enol form of acetaldehyde, vinyl

alcohol, suggest that its decomposition processes are likely to

lie at higher energy than the CC bond ﬁssion of acetone. For

example, the CO and CC bond ﬁssions of the enol form of

acetone lie 30 and 35 kcal/mol higher in energy than the CC

bond ﬁssion in acetone. Barriers to molecular pathways

(dehydration and other processes) also lie 8−11 kcal/mol

above the energy for CC bond ﬁssion of acetone (R1). Thus,

one expects a rapid equilibration of the keto and enol forms of

acetone, but without the opening of any signiﬁcant new

dissociation pathways.

sensitive to rotational temperature, and the wide range of the

derived branching ratio values between 1% and 10% is largely

driven by the relatively poor determination of the rotational

temperature of ketene. Additionally, the intensity for each

measured rotational transition was assumed to have an

uncertainty of 20%, a conservative estimate for CP

spectrometers.²⁸Consistent with the previous experimental

studies¹⁸and the present kinetic model, we ﬁnd that ketene is a

major acetone pyrolysis product, and we report detection of

the previously unknown productacetaldehyde. We have also

detected propene, propyne, and water as minor products, see

Section 3.6.1. Rotational spectroscopy requires a permanent

electric dipole moment for detection. Thus, this experiment is

blind to the nonpolar species CH₃, CH₄, CH₂CH₂, CH₂CCH₂,

H₂, and H. In addition, CO and CH₂CH₃have no transitions

within the bandwidth of the spectrometer. The unexpected

detection of CH₃CHO as a minor product undetected in prior

experimental studies warrants a mechanistic explanation for its

source.

Prior theoretical studies^81,82of 2-propenol decomposition

are consistent with this assessment. However, back-isomer-

ization of the enol (−R6) forms chemically activated acetone

(CH₃C(O)CH₃*), which can then rapidly dissociate via

CH₂C(OH)CH₃→ (CH₃C(O)CH₃*) → CH₃+ CH₃CO

chemical activation

(R7)

3.3. Theory of Bimolecular Reactions and Secondary

Chemistry in Acetone Pyrolysis. As the thermal decom-

position process occurs, the system becomes seeded with both

methyl radicals and CH₃CO radicals from (R1). Radicals are of

course, highly reactive, and so one should next consider their

fate. The CH₃CO radical is a weakly bound radical that near-

instantaneously dissociates to CH₃+ CO at high temper-

atures.⁹¹Thus, we limit our attention to the methyl radical.

This radical may react with acetone via either H abstraction,

This competition between stabilization to acetone (−R6) and

subsequent dissociation of chemically activated acetone to

radical products via (R7) was not characterized in the prior

theoretical studies.^81,82Master equation calculations were

performed with the VRC-TST results for the CC ﬁssion

channel from Saxena et al.⁸³for (R1) coupled with TST

predictions for (R6) using the energies (and molecular

parameters) from the present CCSD(T)/CBS//M06-2X/cc-

pVTZ calculations. Figure S7 in the Supporting Information

summarizes the results from these master equation calculations

performed over the 800−2000 K T-range and pressures

ranging from 0.1 to 10 atm. To summarize, (R1), (R6), and

(R7) are the relevant major channels of interest to acetone

decomposition and the results for (R1) from Saxena et al. and

the present results for (R6) and (R7) are used in the present

kinetics model (see Supporting Information for the entire

model).

3.2. Experimental Observation of Ketene and

Acetaldehyde. The spectrum of 0.5% acetone in argon at a

wall temperature T_wall= 1800 K, and stagnation pressure p₀=

550 mbar, is shown in Figure 1 (top panel). In this article we

report the maximum temperature of the outer wall of the SiC

microreactor, T_wall, as measured by a thermal imager⁶⁰and the

stagnation (inlet) pressure measured upstream from the

microreactor by a capacitance manometer. We emphasize

that the distribution of pressure and temperature inside the

microreactor is nonuniform,⁵⁷⁻⁵⁹and therefore T_walland p₀are

only upper bounds of the T, p distribution. After 2 h of

averaging, the signal-to-noise (S/N) of the strongest transition,

which belongs to acetone, is about 7500:1. After removing all

transitions belonging to acetone (and its isotopologues), the

residuals are shown in the middle panel of Figure 1. The

residuals are dominated by ketene (CH₂CO) transitions, and

CH₃+ CH₃C(O)CH₃→ CH₂C(O)CH₃+ CH₄

(R8)

H abstraction

or addition to either side of the π-bond.

CH₃+ CH₃C(O)CH₃→ (CH₃)₃CO

CH₃addition to C

(R9)

CH₃+ CH₃C(O)CH₃→ (CH₃)₂COCH₃

CH₃addition to O

(R10)

In principle, although the addition can occur at either the C

site or the O site, the C site has a much lower barrier and is

expected to dominate. To complete this picture, the

decompositions of (CH₃)₃CO and (CH₃)₂COCH₃were

studied theoretically.

A schematic diagram of the PES for the reaction of CH₃with

acetone is provided in Figure 2, together with a related PES for

the reaction of CH₃with 2-propenol. For the addition to

acetone, both adducts are expected to simply decompose back

to acetone + methyl radical. The ab initio TST based master

equation predictions for the rates of reactions occurring on this

PES are provided within the mechanism in the Supporting

Information.

Notably, the barrier for abstraction to form CH₃C(O)CH₂is

found to be nearly identical with the lower of the two addition

barriers. The CH₃C(O)CH₂abstraction product, formed in

reaction (R8), is expected to dissociate primarily to CH₂CO +

CH₃.

the rotational temperature of ketene is determined to be T_rot

=

15 5 K.

After removal of CH₂CO (ref 87) lines, a second spectrum

becomes apparent, and is shown in the bottom panel of Figure

1. The carrier is identiﬁed as acetaldehyde (CH₃CHO) with a

rotational temperature T_rot= 5 2 K. Using the PGOPHER

software⁸⁸with the known dipole moments of both

species,^89,90and accounting for the splitting of acetaldehyde

lines due to the CH₃internal rotor levels interactions, the

CH₃₊C_0.H₀₇O:CH₂CO relative abundance is determined to be

0.03_−0.02. The determination of the relative abundance is

CH₂C(O)CH₃→ CH₂CO + CH₃

(R11)

beta scission

Ab initio TST-based master equation rate predictions for

this process were also obtained here. Notably, this radical is a

resonantly stabilized radical with a bond strength of ∼40 kcal/

3129

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

H + CH₃C(O)CH₃→ CH₂C(O)CH₃+ H₂

(R16)

H abstraction

and addition reactions:

H + CH₃C(O)CH₃→ CH₃CH(O)CH₃

(R17)

(R18)

H addition to C

H + CH₃C(O)CH₃→ CH₃C(OH)CH₃

H addition to O

A schematic diagram of the PES for the reaction of H atoms

with acetone and with 2-propenol is provided in Figure 3.

Figure 2. Schematic reaction path diagram for the reaction of CH₃

with acetone [CH₃C(O)CH₃] and its enol isomer [CH₂C(OH)CH₃]

obtained via CCSD(T)-F12/cc-pVTZ-F12//ωB97XD/cc-pVTZ cal-

culations. The primary paths for addition to the terminal and central

carbons in the double bond of the enol are denoted with dark blue

long-dashed and red solid lines, respectively. The dotted light blue

lines denote abstraction paths, while all other paths are denoted with

black dashed lines.

mol. Thus, as indicated by the predicted pressure dependent

kinetics for (R11) (shown in Figure S8), CH₂C(O)CH₃

survives even at quite high temperatures. The combination

of (R8) and (R11) then yield CH₄and CH₂CO as major stable

products. They also recycle the methyl radicals. It is important

to consider what else may happen to the growing pool of

methyl radicals as the pyrolysis proceeds.

The continuing thermal dissociation of the acetone fuel

produces ever more methyl radicals. As its concentration

accumulates, the self-reaction of methyl radicals grows in

importance. This reaction primarily yields ethane molecules:

Figure 3. Schematic reaction path diagram for the reaction of H with

acetone [CH₃C(O)CH₃] and its enol isomer [CH₂C(OH)CH₃]

obtained via CCSD(T)-F12/CBS(TZ-F12,QZ-F12)//B2PLYP-D3/

cc-pVTZ calculations. The lowest energy path for acetone (denoted

by the purple dashed-dotted line) produces acetaldehyde (CH₃CHO)

+ H. The lowest energy path for the reaction with the enol is denoted

with the red solid line. The long-dashed dark blue line denotes the

lowest energy path for addition of H to the O site of the enol. It yields

OH + propene (CH₂CHCH₃). The dotted light blue lines denote

abstraction paths. All other paths are shown as black dashed lines.

CH₃+ CH₃→ C₂H₆

CH₃recombination

(R12)

However, at higher temperature there is also signiﬁcant

branching to ethyl and H radicals (via chemically activated

C₂H₆):⁹²

To complete this picture, the subsequent decompositions of

CH₃CH(O)CH₃,CH₃C(OH)CH₃, and CH₂C(O)CH₃were

also studied theoretically. The predicted rates are provided

within the mechanism in the Supporting Information. These

calculations suggest that the decompositions of the radical

formed in (R16) exclusively forms CH₂CO through β-scission.

Meanwhile, CH₃CHO is formed by the near-instantaneous

dissociation of the weakly bound CH₃CH(O)CH₃radical in

(R17). In the kinetics model this sequence of (R17) and

subsequent rapid dissociation of CH₃CH(O)CH₃is adequately

represented as the direct well-skipping homolytic substitution

reaction

CH₃+ CH₃→ CH₃CH₂+ H

(R13)

addition − elimination

Alternatively, the abstraction of an H atom from ethane also

yields ethyl adicals:

H/CH₃+ C₂H₆→ CH₃CH₂+ H₂/CH₄

(R14)

abstraction

The ethyl radical rapidly dissociates to produce H atoms and

ethylene:

H + CH₃C(O)CH₃→ CH₃CHO + CH₃

CH₃CH₂→ CH₂CH₂+ H

well‐skipping substitution

(R19)

(R15)

beta scission

Thus, far we have considered the reactions of the CH₃and H

radicals with only the keto form of the fuel. Equilibrium

constants for the keto−enol isomerization process calculated

using the ATcT thermochemistry methods are depicted in

Figure 4. At high temperatures the equilibration is expected to

yield a signiﬁcant amount of the enol as was observed in our

prior study of the analogous tautomerization in CH₃CHO.⁴⁰

The net result of the (R12) to (R15) system is to convert CH₃

radicals to C₂hydrocarbons (C₂H₆and C₂H₄) and H atoms

together with H₂and CH₄.

Now the H atoms can participate in reactions with the fuel

in much the same way as CH₃radicals. That is, there are both

abstraction

3130

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

CH₃+ CH₂C(OH)CH₃→ CH₂C(OH)CH₂+ CH₄

H abstraction

(R25)

and addition reactions:

H + CH₂C(OH)CH₃→ CH₃C(OH)CH₃

(R26)

(R27)

(R28)

terminal C addition

H + CH₂C(OH)CH₃→ CH₂CH(OH)CH₃

central C addition

CH₃+ CH₂C(OH)CH₃→ CH₃CH₂C(OH)CH₃

terminal C addition

CH₃+ CH₂C(OH)CH₃→ CH₂C(CH₃)(OH)CH₃

central C addition

(R29)

as illustrated in Figures 2 and 3.

Figure 4. Equilibrium constants for keto−enol processes. The solid

line denotes the present ATcT values for the acetone/2-propenol

equilibrium constant, while the orange dashed line denotes the values

for the acetone/vinyl alcohol equilibrium constant as predicted in ref

40.

The CHC(OH)CH₃and CH₂C(OH)CH₂species are

metastable (although the latter is resonantly stabilized), and

their beta-scissions should immediately yield OH and propyne

or allene.

CHC(OH)CH₃→ CHCCH₃+ OH

(R30)

beta scission

This is evident from the calculated equilibrium constants

shown in Figure 4, where the CH₃C(O)CH₃↔ enol

equilibrium constants are smaller than the CH₃CHO ↔ enol

equilibrium constants by only a factor of 2 (or less) at the high

temperatures (∼1500 K) accessed in the present experiments.

In contrast to the present predictions, a recent study by

Grajales-Gonzalez et al.⁹³concluded that the enol form is

irrelevant to the kinetics of acetone decomposition on the basis

of their equilibrium constants and that their results invalidated

using analogies for keto−enol tautomerizations based on vinyl

alcohol ↔ acetaldehyde. The equilibrium constants from the

Grajales-Gonzalez et al.⁹³study were obtained as the ratio of

their theoretical predictions for k₅and k₋₅from multistructural

variational TST calculations and diﬀer from the equilibrium

constants calculated in Figure 4 by more than 10 orders of

magnitude. This discrepancy is largely due to the theoretically

predicted rate constants for k₅by Grajales-Gonzalez et al.⁹³

that are ∼10 orders of magnitude lower than the present

theoretical predictions. We recommend that Grajales-Gonzalez

et al. revisit their simulations for tert- and 2-butanol pyrolysis

with the present equilibrium constants.

CH₂C(OH)CH₂→ CH₂CCH₂+ OH

(R31)

beta scission

Reactions (R21) and (R24) along with the subsequent rapid

beta-scission (R30) are now adequately represented as

abstractions coupled with prompt dissociation:

H/CH₃+ CH₂C(OH)CH₃→ CHCCH₃+ OH

+ H₂/CH₄

abstraction − dissociation

whereas (R22) and (R25) and the subsequent rapid beta-

scission (R31) are similarly represented as

(R32)

H/CH₃+ CH₂C(OH)CH₂→ CH₂CCH₂+ OH

+ H₂/CH₄

abstraction − dissociation

(R33)

In a similar vein, the radical CH₂CH(OH)CH₃formed from

central C-addition in (R27) exclusively dissociates to OH +

propene and the corresponding well-skipping homolytic

substitution reaction is

The CH₃and H atoms should react with this enol form

through further abstraction reactions:

H + CH₂C(OH)CH₃→ CH₂C(O)CH₃+ H₂

H + CH₂C(OH)CH₃→ CH₂CHCH₃+ OH

(R20)

(R21)

H abstraction

well‐skipping substitution

(R34)

H + CH₂C(OH)CH₃→ CHC(OH)CH₃+ H₂

The decomposition of the addition products was again

explored with theoretical kinetics calculations. The OH formed

in reactions (R31)−(R34) can abstract H atoms from the

acetone and enol to continue the chain as depicted for H

atoms and CH₃radicals, while producing H₂O as a side

product. Related automated kinetics calculations were also

performed for reactions on the C₃H₇O₂surface arising from

the addition of OH with acetone and its enol isomer.

H abstraction

H + CH₂C(OH)CH₃→ CH₂C(OH)CH₂+ H₂

(R22)

(R23)

H abstraction

CH₃+ CH₂C(OH)CH₃→ CH₂C(O)CH₃+ CH₄

H abstraction

CH₃+ CH₂C(OH)CH₃→ CHC(OH)CH₃+ CH₄

An alternative path to propene begins with the addition of

CH₃to C₂H₄with allene and propyne resulting from

subsequent reactions of C₃H₇. This reaction sequence had

H abstraction

(R24)

3131

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Table 2. Acetone Pyrolysis: Major Reactions and Their Thermochemistry from ATcT

Δ_rH°₀

Δ_rH°₂₉₈

Δ_rH°₁₅₀₀

Δ_rH°₁₈₀₀

Δ_rG°₂₉₈

Δ_rG°₁₅₀₀

Δ_rG°₁₈₀₀

kcal/mol

kcal/mol kcal/mol kcal/mol

kcal/mol

CH₃C(O) CH₃→ CH₃CO + CH₃

CH₃C(O)CH₃→ CH₃C(O)CH₂+ H

CH₃C(O)CH₃→ CH₂CO + CH₄

CH₃C(O)CH₃→ CH₃CH₃+ CO

CH₃C(O)CH₃→ CH₃CHO + CH₂

CH₃C(O)CH₃→ CH₂C(OH)CH₃

CH₃C(OH)CH₂→ CH₃CO + CH₃

CH₃CO → CH₃+ CO

CH₃+ CH₃C(O)CH₃→ CH₄+ CH₂C(O)CH₃

CH₂C(O)CH₃→ CH₂CO + CH₃

CH₃+ CH₃→ CH₃CH₃

CH₃+ CH₃→ CH₃CH₂+ H

H + CH₃CH₃→ CH₃CH₂+ H₂

CH₃+ CH₃CH₃→ CH₃CH₂+ CH₄

CH₃CH₂→ CH₂CH₂+ H

H + CH₃C(O)CH₃→ CH₂C(O)CH₃+ H₂

H + CH₃C(O)CH₃→ CH₃CH(O)CH₃

CH₃CH(O)CH₃→ CH₃CHO + CH₃

H + CH₃C(O)CH₃→ CH₃C(OH)CH₃

H + CH₃C(O)CH₃→ CH₃CHO + CH₃

H + CH₂C(OH)CH₃→ CH₂C(O)CH₃+ H₂

H + CH₂C(OH)CH₃→ CHC(OH)CH₃+ H₂

H + CH₂C(OH)CH₃→ CH₂C(OH)CH₂+ H₂

CH₃+ CH₂C(OH)CH₃→ CH₂C(O)CH₃+ CH₄

CH₃+ CH₂C(OH)CH₃→ CHC(OH)CH₃+ CH₄

CH₃+ CH₂C(OH)CH₃→ CH₂C(OH)CH₂+ CH₄

H + CH₂C(OH)CH₃→ CH₃C(OH)CH₃

CHC(OH)CH₃→ CHCCH₃+ OH

CH₂C(OH)CH₂→ CH₂CCH₂+ OH

H + CH₂C(OH)CH₃→ CHCCH₃+ OH + H₂

CH₃+ CH₂C(OH)CH₃→ CHCCH₃+ OH + CH₄

H + CH₂C(OH)CH₃→ CH₂CCH₂+ OH + H₂

CH₃+ CH₂C(OH)CH₃→ CH₂CCH₂+ OH + CH₄

H + CH₂C(OH)CH₃→ CH₂CHCH₃+ OH

OH + CH₃C(O)CH₃→ CH₂C(O)CH₃+ H₂O

CH₃+ CH₂CO → CH₃CH₂+ CO

82.80

95.01

21.05

4.26

104.21

11.63

71.17

9.44

−8.35

29.40

−87.99

11.34

−3.94

−4.03

34.84

−8.26

−9.84

4.77

−22.55

−5.07

−19.88

8.81

−14.52

−19.97

8.72

−14.61

−34.18

30.68

55.15

39.50

39.40

40.63

40.54

1.78

84.46 0.09

96.28 0.16

22.47 0.07

5.36 0.07

83.04

100.26

22.94

4.00

106.97

14.00

69.04

8.97

−6.32

29.26

−88.01

15.10

−4.36

−3.46

38.27

−7.22

−3.21

−5.35

−23.09

−8.56

−21.22

10.29

−13.12

−20.32

11.19

−12.22

−37.10

26.36

82.03

100.66

23.28

3.32

107.18

14.22

67.81

7.39

−5.15

28.44

−86.09

17.03

−4.97

−2.69

38.47

−7.44

0.44

−10.22

−22.71

−9.78

−21.66

9.64

−12.87

−19.38

11.92

−10.58

−36.93

24.82

52.46

34.46

36.74

39.59

41.88

−0.54

−21.88

−22.65

−1.60

72.46

87.83

12.88

−4.12

94.14

11.81

60.66

2.19

−8.37

21.24

−78.77

12.93

−5.47

−4.50

29.45

−9.34

−3.84

−4.48

−17.21

−8.32

−21.14

7.69

23.27

48.47

−27.26

−42.16

44.54

11.41

38.07

−37.32

−51.33

32.04

(R1)

(R2)

(R3)

(R4)

(R5)

(R6)

(R7a)

(R7b)

(R8)

105.88 0.08

11.66 0.16

72.80 0.18

10.98 0.07

−8.64 0.16

31.11 0.17

−90.09 0.04

10.79 0.06

−3.33 0.06

−4.04 0.06

35.92 0.06

−7.93 0.16

−10.83 0.17

6.01 0.19

−23.33 0.20

−4.82 0.08

−19.58 0.21

9.65 0.20

−14.19 0.30

−20.30 0.21

8.94 0.20

−14.90 0.30

−34.99 0.24

31.75 0.22

56.64 0.28

41.39 0.17

40.68 0.17

42.45 0.17

41.74 0.17

1.84 0.17

9.21

14.06

8.23

3.17

−33.96

−8.33

−18.93

−31.47

19.49

−14.30

−5.84

0.21

−16.79

19.84

−39.59

7.74

−19.75

−26.00

−3.24

−22.96

−17.54

5.22

−14.50

−1.47

−9.56

11.08

−12.79

−4.34

−11.88

−3.42

−10.85

−26.95

−20.75

−5.71

−42.40

−8.84

−28.49

−20.34

20.19

−16.24

−6.39

−7.42

−18.69

24.12

(R11)

(R12)

(R13)

(R14a)

(R14b)

(R15)

(R16)

(R17a)

(R17b)

(R18)

(R19)

(R20)

(R21)

(R22)

(R23)

(R24)

(R25)

(R26)

(R30)

(R31)

(R32a)

(R32b)

(R33a)

(R33b)

(R34)

(R35)

(R36)

(R37)

−45.99

13.88

−21.87

−26.92

−5.88

−24.95

−17.07

3.96

−15.10

5.64

−16.60

2.68

−22.48

−12.63

−22.27

−12.42

−12.90

−27.97

−20.40

−6.48

−15.61

−20.17

8.66

−14.64

−29.02

22.59

47.26

30.29

31.26

31.65

32.62

−1.35

−23.22

−21.49

−2.44

53.61

36.64

37.54

40.50

41.40

−0.63

−21.87

−22.39

−2.11

−22.63

−20.83

−0.87

−22.58 0.16

−21.15 0.07

−1.03 0.19

CH₃+ CH₂C(O)CH₃→ CH₃CH₂+ CH₃CO

been assumed to be the primary source of propene in prior

modeling studies.^15,23However, the modeling results described

below demonstrate that enol-driven chemistry is the primary

source for C₃H₆, allene, and propyne at the conditions of

relevance to the present study.

For reference, the ATcT predictions for the enthalpy and

free energy changes in the main reactions considered here are

summarized in Table 2. This table includes 0 and 298.15 K

values, as well as values both at the nominal (T_wall= 1800 K)

temperature of the experiment and at 1500 K, the likely

temperature at the reactor centerline (vide infra). Two

conformers of 2-propenol, CH₂C(OH)CH₃, and four con-

formers of CHC(OH)CH₃are characterized in Figure S9 and

S10, respectively.

3.4. The Kinetic Model. 3.4.1. Branching between the

Substitution and Abstraction Pathways. Kinetics simulations

with the present acetone model relevant to the current

experimental observations are presented in Figure 5. The

boundary layer model from earlier work⁵⁷concludes that the

ﬂow within the SiC tube with Ar as the bath gas can cause axial

and radial distributions with signiﬁcant gradients in the state

variables and species concentrations and precludes the use of a

plug-ﬂow model. Therefore, we have used the results from the

boundary-layer model to guide the homogeneous, adiabatic

reactor simulations presented in Figure 5. Those results⁵⁷

indicate that, with Ar as the bath gas, the centerline

temperature of the microreactor is 200−300 K colder than

the wall temperature along with a pressure decrease (ca. 0.5p₀)

along the axial direction of the SiC tube. The simulations

shown in Figure 5 were therefore performed at pressures of

0.25 bar (p₀= 0.55 bar) with [CH₃C(O)CH₃]₀= 0.5% in Ar as

the bath gas. The simulated ratios of [CH₃CHO]/[CH₂CO]

using the present kinetics model show good agreement with

the experimentally derived ratio for T = 1500 K (representing a

300 K temperature decrease from T_wall= 1800 K) and at t = 50

μs residence time. These conditions are in good agreement

with the mean temperatures and residence times observed

using the boundary layer model from our previous study and

allows us to use these conditions to analyze the reaction

pathways responsible for the observed intermediates in the

present study. For completeness, Figures S11 −S14 depict the

simulated time evolution of the molecules detected in the

present experiments at the high temperatures relevant to the

SiC microreactor.

3132

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Figure 5. Present closed, homogeneous reactor simulations of the

acetaldehyde to ketene concentration ratios for [CH₃C(O)CH₃]₀=

0.5% in Ar at initial temperatures ranging from 1200−1800 K are

represented by colored solid lines. The open square symbol denotes

the present experimentally derived branching ratio along with its

associated error bars.

Figure 6. Plot of the temperature dependence of the high-pressure

limit thermal dissociation rate constants for CH₃CH(O)CH₃. In this

calculation it is assumed that the CH₃CH(O)CH₃is thermalized,

which we do not expect to be the case in our experiment.

Nevertheless, it follows from these rates that an upper bound for the

lifetime of the chemically activated CH₃CH(O)CH₃complex in

(R19) is ∼1 ps at the present experimental conditions. The inset

shows the equilibrium geometry of the complex.

Figure S15 depicts Arrhenius plots for the present

theoretically predicted rate constants for the two main

channels in the reaction of H atoms with CH₃C(O)CH₃,

(R16) abstraction and (R19) homolytic substitution. If the

only formation/rate-limiting processes for CH₂CO and

CH₃CHO are (R16) and (R19) respectively, then the ratio

k₁₈/k₁₅≈ 0.14 (at 1500 K) can be a crude estimate for the

CH₃CHO to CH₂CO concentration ratios. However, with the

plethora of unimolecular and bimolecular reactions that

consume both species at high temperatures, the present

well-skipping in the combined substitution reaction (R19) at

the high temperatures of the present experiment.

Figure S16 in the Supporting Information illustrates the

results for the simulated CH₃CHO to CH₂CO concentration

ratios at the same conditions as in Figure 5, but without the

inclusion of the direct process (R19). The exclusion of (R19)

reduces the simulated concentration ratios by orders of

magnitude emphasizing the critical role of the well-skipping

reaction (R19) in simulating the formation of CH₃CHO. It is

perhaps worth noting though, if the well-skipping ﬂux was

instead attributed to the addition process (R17), rather than

simply being removed, then the two simulations would be in

much better agreement.

experimentally derived concentrations ratio [CH₃CHO]/

+0.07

−0.02

[CH₂CO] = 0.03

is signiﬁcantly smaller than the 0.14

reaction rates ratio. This measured branching ratio is in

agreement with the 1500 K predictions of the kinetic

simulations as shown in Figure 5.

3.4.2. Well-Skipping Substitution Kinetics. As mentioned

in the introduction, the formation of CH₃CHO is not a

consequence of the sequential process involving reaction

(R17) and the subsequent β-bond scission of the product of

(R17), CH₃CH(O)CH₃. Instead, the observed CH₃CHO is

entirely due to the direct process (R19). The well-skipping

nature of this process can be understood from the plot in

Figure 6 of the high-pressure limit¹¹¹thermal dissociation rate

constants for CH₃CH(O)CH₃. In the Gedankenexperiment of

Figure 6, the CH₃CH(O)CH₃complex is considered to be

thermalized at a temperature shown on the abscissa. Even if

such a hypothetical situation could be achieved, the lifetime of

the thermally equilibrated complex at 1500 K (52.89 kcal/mol

of internal energy) would be on the order of 1 ps. In the

present experiment though, this complex is formed via (R19)

and thus possesses additional internal energy. As a

consequence, its lifetime is most likely even shorter than 1

ps, which is already far too short for collisional stabilization to

occur. Furthermore, the presence of the energetically accessible

exit channel, CH₃CHO + CH₃, facilitates rapid reaction to

these products rather than back-dissociation to reactants

(Figure 3). The 14 kcal/mol barrier to that exit channel is

deeply submerged at these internal energies, which facilitates

3.4.3. Path Flux Analysis. Figure 7 depicts the results from

the present kinetics model at p = 0.25 bar and T = 1500 K for

the key products observed in the present experimental study.

The proﬁle for 2-propenol is also included in Figure 7 because

of its prominent role in driving the formation of the observed

intermediates. The simulations predict ∼10% extent of

decomposition of the reactant acetone at 50 μs with noticeable

amounts of the dominant product CH₂CO being produced.

The simulations predict lesser amounts of CH₃CHO, H₂O,

CH₃CHCH₂, and CH₃CCH being produced in mole fractions

<10 ppm under these conditions. Notably, the 2-propenol is

predicted to rapidly accumulate to mole fractions of ∼10⁻⁴.

Figure S17 in the Supporting Information presents the

evolution of the other major and minor intermediates at the

conditions of the present experiments, p = 0.25 bar and T =

1500 K using the present model.

In order to understand the mechanistic pathways to

products, path ﬂux analyses have been performed.⁹⁴The

species concentrations are obtained by solving coupled

diﬀerential equations. The net chemical production rate of

each species results from a competition between the ﬂuxes of

all the chemical reactions involving that species, and

subsequent analyses of these instantaneous ﬂuxes is used to

3133

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Figure 8. Path ﬂux analysis using the present acetone kinetics model.

The ﬂux analysis has been performed at the same initial conditions as

in Figure 7 at t = 50 μs. The species detected in the present

experiment are highlighted with red font and circles.

both acetone precursors and substitute the CH₃and CD₃

groups as in (R19):

CH₃C(O)CH₃+ H → CH₃CHO + CH₃

CH₃C(O)CH₃+ D → CH₃CDO + CH₃

CD₃C(O)CD₃+ H → CD₃CHO + CD₃

CD₃C(O)CD₃+ D → CD₃CDO + CD₃

Figure 7. Closed, homogeneous reactor simulations illustrating the

species mole fractions predicted for [CH₃C(O)CH₃]₀= 0.5% in Ar at

T = 1500 K and p = 0.25 bar. The species detected in the present

experiments as well as the key intermediate 2-propenol are illustrated

with color lines. The species in the legend are listed in the order of

increasing (in the direction of the arrow) concentration at the

experimental residence time of 50 μs (marked with the vertical dashed

line).

As shown in Figure 9, each of the isotopologuesCH₃CHO,

CH₃CDO, CD₃CHO, and CD₃CDO (refs 31, 95−97)is

observed with roughly equal intensity, which is consistent with

this mechanism. A possible alternative explanation for the

observed products would be the existence of some H−D

scrambling in the hot reactor. However, if that were the

mechanism leading to the observed products, we would expect

to also observe the acetaldehyde species with mixed methyl

rotors CH₂D and CD₂H,^98,99where a free H or D atom

substitutes into the methyl group. None of those species are

observed.

The four observed isotopologues of acetaldehyde conﬁrm

that a bimolecular mechanism is responsible for acetaldehyde

formation in the pyrolysis of acetone. Indeed, if acetaldehyde

was a result of unimolecular elimination (R5), the “crossover”

products CH₃CDO and CD₃CHO would be absent. This

experiment with isotopically substituted acetone does not by

itself rule out a hypothetical chemical pathway in which acetyl

radicals lead to acetaldehyde via self-reaction or abstraction of

an H atom. In order to evaluate that possibility, we rely on the

modeling component of our MET framework. The kinetic

model described in Section 3.4. rejects the signiﬁcance of that

acetyl-mediated pathway based on the transient nature of this

weakly bound⁹¹radical.

generate the path ﬂux diagram shown in Figure 8. The ﬂux

analysis is performed at the initial conditions shown in Figure 7

at t = 50 μs. Intermediates observed in the present

experimental studies are circled and highlighted in red.

Reported values are relative rates of consumption for the

various reactants in the ﬂux diagram except for H₂O, where the

values represent relative rates of formation.

Unimolecular reactions (R1) and (R2) are responsible for a

large fraction (∼80%) of acetone consumption, with

bimolecular reactions with H and CH₃accounting for the

remainder. The major product ketene originates predom-

inantly from the beta-scission of the acetonyl radical

CH₂C(O)CH₃via (R11) with the source of acetonyl being

the hydrogen abstraction reactions (R8) and (R16) from

acetone. CH₃CHO formation is nearly exclusively (>96%) due

to the well-skipping homolytic substitution reaction (R19),

with the kinetics for this reaction calculated using the

automated kinetics code.

3.5. Isotopic Experiment. In order to experimentally

conﬁrm the bimolecular nature of the acetaldehyde formation

mechanism, the spectrum of the pyrolysis products of a 1:1

CH₃C(O)CH₃:CD₃C(O)CD₃mixture was recorded. For this

“crossover” experiment, we expect the mechanism below to

generate several H, D isotopologues of acetaldehyde in roughly

equal abundance. The initial pyrolysis steps are analogous to

reactions (R1, R13, R15). In this deuterated experiment, they

produce equal amounts of H and D atoms, which then attack

3.6. The Role of Enol Chemistry. In addition to the

pathway involving substitution of CH₃by H atom, there is also

evidence for participation by the enol form of acetone. This

stems from the detection of propene, propyne, and H₂O lines

in the present experiment, all of which are attributed to enol

driven chemistry. CH₃C(O)CH₃is converted via (R6) to the

enol (the syn or the anti rotamer of 2-propenol, CH₂C(OH)-

CH₃, see Figure S9) by isomerization. Subsequently, the enol

3134

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Figure 9. Rotational spectra of deuterated acetaldehyde products in the pyrolysis of acetone isotopologues. Detection of CH₃CDO and CD₃CHO

conﬁrms that it is the bimolecular chemistry that is responsible for acetaldehyde formation in the pyrolysis of acetone. In this experiment, the 1:1

mix of acetone:acetone-d₆, the total concentration 1% by mole in argon, is ﬂown through the T_wall= 1800 K microreactor, and, along with the

reaction products, adiabatically cooled in the supersonic expansion for millimeter-wave detection. (Top Panel) The black trace is the 75−110 GHz

pyrolysis spectrum (T_wall= 1800 K) after averaging 5 × 10⁵FIDs from each 30-MHz segment. (Bottom Panels) magniﬁcations highlighting the

4₀₄−3₀₃transitions of CH₃CHO and CH₃CDO, and magniﬁcations highlighting the 5₀₅−4₀₄transitions of CD₃CHO and CD₃CDO.

could react with a free H atom to form propyne (CH₃CCH)

via the H atom abstraction reaction (R21). Alternatively,

propene (CH₂CHCH₃) may be formed by addition of H to 2-

propenol followed by elimination of OH (R34); another

homolytic substitution reaction similar to reaction (R19). The

present path ﬂux analysis in Figure 8 indicates that the well-

skipping substitution reaction (R34) is the dominant source

for the formation of propene.

3.6.1. Experimental Observation of the Enol-Mediated

Products. Propyne¹⁰⁰is observed in this work as a minor

product. Propyne was also detected in a previous study²⁵but a

compelling explanation for its presence was lacking. Propyne

line intensities were found to be highly dependent on the

reactant concentration and their detection required a

minimum acetone concentration of 10% (T_wall= 1800 K) for

observation at 3:1 S/N in a broadband spectrum, as shown in

Figure 10. The simulations in Figure 7 indicate that <1 ppm of

propyne is expected when [CH₃C(O)CH₃]₀= 0.5%. Path ﬂux

analysis indicates that the enol driven reactions represented by

the abstraction reaction (R32) are the source for the observed

propyne.

analyses presented in Figure 8 indicate that reaction (R34) is

the primary source not only for propene but also for its

associated coproduct OH in this reaction. Reaction of OH

with acetone is expected to be facile and the ﬂux analysis in

Figure 8 indeed conﬁrms that abstraction (R35) to form H₂O

is the primary source for its production:

OH + CH₃C(O)CH₃→ CH₂C(O)CH₃+ H₂O

(R35)

abstraction

We ﬁnd that at acetone concentrations above 2%, the

precursor signal is no longer proportional to the acetone

concentration. At very high concentrations (ca. 10%) of

acetone and T_wall= 1800 K, the mixture causes adverse eﬀects

to the reactor by building a residue inside the SiC tube.

Therefore, we do not determine relative abundances for

CH₃CCH, CH₂CHCH₃, and HDO under these conditions.

Allene (CH₂CCH₂) is devoid of a permanent dipole moment

and H₂O and OH are not detectable within the spectrometer’s

bandwidth, so we cannot conﬁrm their presence.

A spectroscopic search for both conformers of the enol

(Figure S9) form of acetone was undertaken using rotational

constant and dipole moment predictions at the B2PLYP-D3/

cc-pVTZ level of theory^103−105for both equilibrium (r_e) and

vibrationally averaged ground-state (r₀) structures. However,

no spectra consistent with either enol conformer were

detected. The lowest energy conformer of the enol form, syn,

is stabilized by ΔE₀= 1.5 kcal/mol = 750 K relative to the anti

form, but has a signiﬁcantly smaller μ_adipole component of 0.5

D (compared to 2.1 D). Thus, if the enol exists postpyrolysis

and is appreciably cooled, it may be diﬃcult to identify since

the lowest energy syn conformer has signiﬁcantly weaker

Propene is detected here in trace amounts. In order to detect

propene the spectrometer had to be set up to acquire a

narrowband spectrum around the two known 5₀₅−4₀₄

rotational transitions¹⁰¹at elevated concentration of the

precursor (Figure 10).

The 1₁₀−1₁₁line¹⁰²of HDO is detected in the pyrolysis of a

CH₃C(O)CH₃:CD₃C(O)CD₃mixture diluted in argon at 1%

by mole (Figure 10). Because we only detect a single transition

of HDO, its rotational temperature, a crucial parameter in

deducing abundances, cannot be measured. The path ﬂux

3135

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

barrier that separates the syn and anti rotamers. Therefore, the

hindered torsion separating these conformers is better

represented as a large amplitude motion above the barrier;

there are no distinct conformations. However, as the gas is

passing the resistively heated section of the microreactor

between the two electrodes and enters the terminal 3 mm of

the SiC tube, the temperature falls by 200−500 K. The enol

begins stabilizing into its syn rotamer. Likewise, the total

equilibrium concentration of enol ([syn enol] + [anti enol]) is

expected to fall in the colder exit channel of the reactor

according to Figure S11. The ﬁnal distribution of the syn enol,

anti enol, and acetone forms of C₃H₆O is reached at some

stage of the adiabatic expansion and is diﬃcult to evaluate.

These qualitative considerations reconcile the upper bound

ratios derived in the previous paragraph with the theoretical

predictions for the enol molar fractions of 2.3 × 10⁻⁸, 3.8 ×

10⁻⁶, 7.0 × 10⁻⁵, 7.6 × 10⁻⁵, 2.2 × 10⁻⁶at 1000, 1200, 1400,

1600, and 1800 K, respectively (Figure S11). These values

correspond to the 5 × 10⁻³initial mole fraction of acetone and

50 μs residence time.

A recent study by Couch et al.¹⁰⁸used a novel tabletop VUV

light source to detect the elusive enol intermediate from

acetone pyrolysis at high temperatures (1000−1400 K). Couch

et al. observe noticeable decomposition of acetone at 1100 K

with the corresponding formation of 2-propenol. At the highest

temperatures (1400 K) Couch et al. observe as much as 60%

conversion of acetone. Simulations using the present model

indicate that with 1% acetone the extent of conversion of the

reactant is <10% at 1400 K, p = 0.25 bar and at t = 60 μs

(estimated residence time in the SiC microreactor used by

Couch et al.). Although the present simulations are in

substantial disagreement with the temperatures at which 60%

of acetone is dissociated and at which the enol concentration

peaks in the Couch et al.¹⁰⁸experiments, the predicted

behaviors of both species’ concentrations (Figure S11 and the

values in the preceding paragraph) are qualitatively similar to

the trends displayed by the experimental signals in ref 108. The

[CH₂C(OH)CH₃]/[CH₃C(O)CH₃] ratios predicted by the

present model at p = 0.25 bar and at t = 50 μs are 4.6 × 10⁻⁶,

7.7 × 10⁻⁴, 1.6 × 10⁻², 3.4 × 10⁻², 4.6 × 10⁻²at 1000, 1200,

1400, 1600, and 1800 K, respectively. Unfortunately, these

data cannot be directly compared to the results of Couch et al.

because their signal needs to be adjusted for the presumably

diﬀerent ionization cross sections of the ketone and enol in

order to be converted to branching ratios.

3.6.2. Kinetic Simulations of the Enol Contribution. In

order to understand the role of the enol-driven chemistry in

acetone, Figure 11 presents simulations performed using the

present model with (shown as solid lines) and without the

keto−enol tautomerization reaction (R6) (shown as dashed

lines). There is a noticeable diﬀerence in the formation of the

enol, 2-propenol, with its concentrations predicted to be a

factor of 30 lower in the absence of (R6). Consequently,

noticeable diﬀerences are seen in the simulated proﬁles of

propene, propyne, and H₂O all of which were observed in the

present experiments. In order to understand if the enol driven

chemistry is exclusively relevant only in the present studies, we

have also simulated the formation of propene and propyne as

well as other intermediates formed in the reﬂected shock tube

studies of Sato and Hidaka. Figure S18 presents the

simulations performed with the acetone model as also depicted

in Figure S1 in comparison to the experimental data. Also

shown in Figure S18 (dashed lines) are the simulations

Figure 10. (Top Panel) The black trace is the 75−110 GHz pyrolysis

spectrum (T_wall= 1800 K) of 10% acetone in argon by mole after

averaging 10⁶FIDs from each 30-MHz segment. After 2000× vertical

and 1200× horizontal magniﬁcation, the J = 5−4 transitions of

propyne are seen. The blue trace is a T_rot= 10 K simulation of

propyne. (Left Bottom Panel) The portion of the above spectrum

containing the HDO transition. (Right Bottom Panel) A narrowband

(2.5 MHz) spectrum of propene obtained using the Targeted mode of

the BrightSpec spectrometer with 5% acetone in argon by mole and

after averaging 4 × 10⁶FIDs.

transitions than its anti counterpart at ﬁxed number density. It

is also diﬃcult to make a reliable prediction for a conforma-

tional temperature by which to estimate the relative

abundances of these two conformers at experimental

conditions, so we have estimated their abundances independ-

ently, and relative to the observed acetone spectrum.

Given that the pyrolysis spectrum of acetone at 5 × 10⁻³

initial mole fraction exhibits a well-deﬁned post-supersonic

expansion rotational temperature of approximately 5 K, we

assume a similar rotational temperature for its enol counter-

part. To calculate an upper bound estimate for the enol

concentration, we calculate the transition frequencies and

intensities of both enol conformers and acetone at a rotational

temperature of 5 K. For the enol, we use the B2PLYP-D3/cc-

pVTZ level of theory^103−105to calculate rotational constants

and dipole moment components.¹¹²SPCAT^106,107was used to

calculate the approximate “rigid rotor” CH₂C(OH)CH₃and

CH₃C(O)CH₃rotational partition functions and to simulate

their spectra at 5 K. A correction factor of 3/6 is introduced to

the ratio of the enol/keto rotational partition functions that

arise from interacting CH₃internal rotor levels in CH₂C-

(OH)CH₃and in CH₃C(O)CH₃. We estimate the upper

bound ratios of [syn enol]/[acetone] ≈ 4 × 10⁻²and [anti

enol]/[acetone] ≈ 2 × 10⁻⁴. These are the relative

concentrations of syn enol and anti enol that we would be

able to detect with our spectrometer in the supersonic

expansion.

At a gas temperature of 1500−1800 K during pyrolysis, the

internal energy of the enol is well above the hydroxyl torsional

3136

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

bond-ﬁssion reaction (R1). These mechanistic processes are

also the same sources for these species as in the literature

kinetics models^15,23,25despite the diﬀerence in kinetic

parameters adopted for these reactions in the present and

literature models.

Interestingly, signiﬁcant mechanistic diﬀerences are ob-

served between the present model and the two recent acetone

models^15,23for sources leading to the formation of C₂H₅and

ultimately C₂H₄from its thermal dissociation. Literature

models predict that the reaction of CH₃radicals (R13) and

hydrogen abstractions from C₂H₆are the only sources for

C₂H₅radicals, whereas the present model includes the

reactions of CH₃radicals with CH₂CO and CH₂C(O)CH₃

(to a lesser extent) as additional minor sources for C₂H₅via

well-skipping substitution processes

CH₃+ CH₂CO → CH₃CH₂+ CO

well‐skipping substitution

(R36)

CH₃+ CH₂C(O)CH₃→ CH₃CH₂+ CH₃CO

Figure 11. Plot of simulated mole fractions assuming a closed,

homogeneous reactor with [CH₃C(O)CH₃]₀= 0.5% in Ar at T =

1500 K and p = 0.25 bar. The solid lines represent this acetone kinetic

model, and the dashed lines correspond to a model excluding the

keto−enol tautomerization reaction.

well‐skipping substitution

(R37)

Reaction (R36) can be thought to occur in an analogous

fashion to the reactions of H atoms with ketene⁹¹by either

addition across the CC π-bond to make CH₃CH₂CO or

CH₂C(O)CH₃or abstraction to make HCCO and CH₄.

Addition to form CH₂C(O)CH₃and related H atom shift

reactions of the acetonyl radical were characterized using the

automated kinetics methods described earlier. The kinetics for

addition to form CH₃CH₂CO, subsequent β-scission of

CH₃CH₂CO to C₂H₅+ CO, and direct abstraction to CH₄

+ HCCO were characterized with master equation calculations

using the energies (and molecular parameters) from the

present CCSD(T)/CBS//M06-2X/cc-pVTZ calculations (see

Table S1 for energetics for the processes). Given the large

exothermicity of the dissociation of the adduct to CH₃CH₂+

CO relative to the CH₃+ CH₂CO entrance channel,

stabilization to CH₃CH₂CO is a negligible process and the

dominant process is well-skipping homolytic substitution

(R36), similar to what was observed⁹¹in the analogous H +

CH₂CO reaction. The model in the Supporting Information

includes the ﬁts to the calculated kinetics for (R36) and the

CH₃+ CH₂CO → CH₄+ HCCO abstraction reaction.

In the experimental studies of Yu et al.¹⁵on acetone

pyrolysis at 1 atm using a jet-stirred reactor at 700−1100 K

methyl ethyl ketone was observed as a minor product and

attributed to the recombination of CH₃radicals with the

resonantly stabilized acetonyl radical. This reaction is also the

source for methyl ethyl ketone in the present model. However,

since C₂H₅+ CH₃CO products are mildly exothermic to the

CH₃+ CH₂C(O)CH₃entrance channel, and CH₃+ C₂H₅CO

is thermoneutral, recombination to methyl ethyl ketone is not

expected to be the only relevant channel at high temperatures.

The kinetics for recombination and addition−elimination in

CH₃+ CH₂C(O)CH₃in the present model (Supporting

Information) was assumed to have the same rate constants as

the analogous recombination and addition−elimination

processes in CH₃+ isobutyl, which were characterized in a

recent study.¹¹⁰

without (R6). While Sato and Hidaka²⁵did not observe the

enol intermediate in their acetone pyrolysis studies, the present

simulations indicate that the enol concentrations can

accumulate to ∼100 ppm at T < 1300 K. Without the

inclusion of (R6) the predicted enol concentrations drop by an

order of magnitude and correspondingly the simulated proﬁles

of propene, propyne, and allene are all observed to be

noticeably lower. We note here that Sato and Hidaka do not

consider the keto−enol tautomerization and the subsequent

chemistry of the enol in their model. The formation of propene

is attributed to CH₃+ C₂H₄and subsequent propyne and

allene formation is attributed to bimolecular reactions of

propene and allyl. A recent experimental and modeling study

by Yu et al.¹⁵arrive at the same mechanistic arguments as

proposed by Sato and Hidaka, essentially ignoring the enol

driven chemistry, which is shown here to be the major direct

contributor to the formation of the C₃hydrocarbon

intermediates. Note that these C₃intermediates likely play

an important role in soot formation pathways through

formation of the propargyl and allyl resonantly stabilized

radicals, the recombination reactions of which commonly

provide the main pathway to formation of the ﬁrst aromatic

ring.¹⁰⁹

Apart from the intermediates detected here, prior high

temperature literature studies^15,25,26indicate that the other

major intermediates that evolve from acetone pyrolysis are

CH₄, CO, C₂H₆, and C₂H₄, which accumulate to mole

fractions of ∼10⁻⁴−10⁻³. Simulations in Figure S17 indicate

that these intermediates are produced in copious amounts even

at the shorter time scales relevant to the present SiC

microreactor studies. Path ﬂux analysis using the present

acetone model indicates, predictably, that the abstraction

(R25) and methyl recombination (R12) reactions are the

major sources for CH₄and for C₂H₆, respectively, in accord

with prior studies. At high temperatures CO originates from

the near-instantaneous dissociation⁹¹of the weakly bound

CH₃CO radical that is predominantly formed in the initiating

Simulations of C₂H₄and CH₂CO formation from literature

studies of acetone pyrolysis by Lam et al.²⁴and Sato and

Hidaka²⁵were performed in order to characterize the roles of

the two C₂H₅formation reactions (R36) and (R37) included

3137

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Table 3. Acetaldehyde Pyrolysis: Major Reactions and Their Thermochemistry from ATcT

Δ_rH°₀

Δ_rH°₁₅₀₀

Δ_rH°₁₈₀₀

Δ_rG°₂₉₈

Δ_rG°₁₅₀₀

Δ_rG°₁₈₀₀

kcal/mol

Δ_rH°₂₉₈kcal/mol

kcal/mol

CH₃CHO → CH₃+ HCO

HCO → H + CO

2HCO → CH₂O + CO

HCO + CH₃CHO → CH₂O + CH₃CO

CH₃CHO + H → CH₂CHO + H₂

CH₂CHO → CH₂CO + H

CH₃CHO + H → CH₂O + CH₃

CH₃CHO + H → CH₃CO + H₂

CH₃CHO → CH₂CHOH

82.77

14.54

−72.17

1.15

−9.06

35.27

−3.95

−15.40

9.96

84.55 0.06

15.71 0.03

−72.48 0.03

1.09 0.06

−8.64 0.15

36.63 0.15

−3.63 0.06

−14.93 0.06

10.00 0.10

82.86

17.71

−73.23

0.66

−7.65

38.25

−8.08

−15.87

11.40

81.42

17.78

−73.29

0.74

−7.76

38.54

−9.65

−16.28

11.34

73.54

9.44

−70.17

1.18

−9.64

29.87

−6.08

−16.38

10.20

28.00

−18.95

−59.76

2.20

−15.83

−0.14

−12.81

−22.24

8.66

17.16

−26.29

−57.06

2.51

−17.46

−7.84

−13.61

−23.47

8.11

in the present acetone model. Figure S19 depicts a simulation

of the time-resolved formation of C₂H₄at high temperatures

from the Lam et al. study. They simulated their C₂H₄data

using the Pichon et al.²³model, which includes only reaction

(R13) as the major source for C₂H₅. Exclusion of reactions

(R36) and (R37) from the present model would reduce C₂H₄

formation, in disagreement with the experimental measure-

ment at 1474 K. The noticeable eﬀect of the inclusion of

reactions (R36) and (R37) is also observed in simulations of

the Sato and Hidaka single-pulse-shock tube measurements of

C₂H₄and CH₂CO presented in Figure S20. The present

combined MET framework has highlighted the prominent

roles of keto−enol tautomerization and well-skipping reactions

that have not been considered in recent literature kinetics

models^15,23to describe the evolution of intermediates from the

high temperature pyrolysis kinetics of acetone.

3.7. The Pyrolysis of Acetaldehyde. Finally, it is

interesting to consider the results of the present study in

relation to the observations from our prior study of the

pyrolysis of acetaldehyde (CH₃CHO) using a pulsed micro-

reactor at 1800 K (ref 36). ATcT predictions for several

relevant reactions are given in Table 3. In that work, the

rotational transitions of ketene (CH₂CO) and formaldehyde

(CH₂O) were observed in a CP-FTmmW spectrum. Although

the substitution reaction can explain CH₂O formation from

CH₃CHO pyrolysis, there may be alternate sources for

formation of acetaldehyde. In particular, HCO is more stable

at high temperatures than CH₃CO and so can accumulate to

signiﬁcant concentrations. Subsequent bimolecular reactions of

HCO with reactant CH₃CHO may be an important source for

CH₂O. Correspondingly, acetone decomposition is a better

chemical system for characterizing the role of such well-

skipping substitutions. The present combined MET framework

adequately establishes the role of these reactions in the

formation of products in the nascent pyrolytic chemistry of

acetone.

model to describe the formation of the numerous inter-

mediates observed in the present and prior studies.

CH₃CHO has not previously been observed as a product in

acetone pyrolysis. The present theory/modeling studies

attribute its formation to a well-skipping homolytic sub-

stitution process that involves addition of an H atom to

acetone followed by the loss of a CH₃group. Keto−enol

tautomerization is also shown to be an important process in

acetone pyrolysis, with enol driven chemistry being the

dominant source for the propene, propyne, and water observed

in the present experiments. Propene is a product of another

substitution reaction in which free H atoms replace OH in the

enol isomer of acetone. In a broader context, this work

emphasizes the importance of the radical substitution reactions

that are often overlooked. The theory-aided model also was

used to simulate the formation of other intermediates in high

temperature shock tube studies from the literature. While the

extended Rice−Herzfeld mechanism was deemed suﬃcient to

describe acetone pyrolysis in studies to date, the present MET

framework highlights the prominent role of enol driven

chemistry and well-skipping reactions to advance our under-

standing of the pyrolytic dissociation of the simplest ketone.

ASSOCIATED CONTENT

* Supporting Information

The Supporting Information is available free of charge at

https://pubs.acs.org/doi/10.1021/jacs.0c11677.

■

sı

Figures S1−S20; Tables S1−S2 (PDF)

Methodological details for automated theoretical

kinetics calculations; Figure S21 (PDF)

Supporting Data S1 (TXT)

Supporting Data S2 (TXT)

Supporting Data S3 (TXT)

AUTHOR INFORMATION

Corresponding Author

■

Kirill Prozument − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States; orcid.org/0000-0003-3045-8429;

Email: prozument@anl.gov

4. CONCLUSION

Broadband rotational spectroscopy was used to quantify

selected products from the high temperature pyrolysis of

acetone. Ketene was observed as the major product, along with

the minor products: acetaldehyde, propyne, propene, and H₂O

(from measurements of HDO in deuterated acetone experi-

ments). An automated theoretical kinetics code was used to

rapidly calculate a priori rate constants for the numerous

addition and abstraction reactions of acetone and its enol

isomer (2-propenol) with H, CH₃, and OH and to model the

subsequent dissociations of radicals formed by substitutions/

abstractions. These results were used to assemble a kinetics

Authors

Daniel P. Zaleski − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States; Department of Chemistry, Colgate

University, Hamilton, New York 13346, United States;

orcid.org/0000-0003-0153-9158

Raghu Sivaramakrishnan − Chemical Sciences and

Engineering Division, Argonne National Laboratory, Lemont,

3138

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Illinois 60439, United States; orcid.org/0000-0002-

1867-1254

Hailey R. Weller − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States; Department of Chemical and

Biomolecular Engineering, Vanderbilt University, Nashville,

Tennessee 37235, United States

Nathan A. Seifert − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States

David H. Bross − Chemical Sciences and Engineering Division,

Argonne National Laboratory, Lemont, Illinois 60439,

United States; orcid.org/0000-0002-8218-0249

Branko Ruscic − Chemical Sciences and Engineering Division,

Argonne National Laboratory, Lemont, Illinois 60439,

United States; orcid.org/0000-0002-4372-6990

Kevin B. Moore, III − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States

The eﬀorts of A.V.C. and S.N.E. research were supported by

the Exascale Computing Project (ECP), Project Number: 17-

SC-20-SC, a collaborative eﬀort of two U.S. Department of

Energy (DOE) organizations, the Oﬃce of Science and the

National Nuclear Security Administration, responsible for the

planning and preparation of a capable exascale ecosystem

including software, applications, hardware, advanced system

engineering, and early test bed platforms to support the

nation’s exascale computing imperative. H.R.W. was supported

by the U.S. Department of Energy, Oﬃce of Science, Oﬃce of

Workforce Development for Teachers and Scientists (WDTS)

under the Science Undergraduate Laboratory Internship

(SULI) program.

REFERENCES

■

(1) Atef, N.; Kukkadapu, G.; Mohamed, S. Y.; Rashidi, M. A.;

Banyon, C.; Mehl, M.; Heufer, K. A.; Nasir, E. F.; Alfazazi, A.; Das, A.

K.; Westbrook, C. K.; Pitz, W. J.; Lu, T.; Farooq, A.; Sung, C.-J.;

Curran, H. J.; Sarathy, S. M. A comprehensive iso-octane combustion

model with improved thermochemistry and chemical kinetics.

Combust. Flame 2017, 178, 111−134.

(2) Zhou, D. D. Y.; Davis, M. J.; Skodje, R. T. Multitarget Global

Sensitivity Analysis of n-Butanol Combustion. J. Phys. Chem. A 2013,

117, 3569−3584.

Sarah N. Elliott − Chemical Sciences and Engineering Division,

Argonne National Laboratory, Lemont, Illinois 60439,

United States

Andreas V. Copan − Emmanuel College, Natural Sciences

Department, Franklin Springs, Georgia 30639, United States

Lawrence B. Harding − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States; orcid.org/0000-0002-5867-4716

Stephen J. Klippenstein − Chemical Sciences and Engineering

Division, Argonne National Laboratory, Lemont, Illinois

60439, United States; orcid.org/0000-0001-6297-9187

Robert W. Field − Department of Chemistry, Massachusetts

Institute of Technology, Cambridge, Massachusetts 02139,

United States; orcid.org/0000-0002-7609-4205

(3) Curran, H. J. Developing detailed chemical kinetic mechanisms

for fuel combustion. Proc. Combust. Inst. 2019, 37, 57−81.

(4) Moore, D. P.; Remedios, J. J.; Waterfall, A. M. Global

distributions of acetone in the upper troposphere from MIPAS

spectra. Atmos. Chem. Phys. 2012, 12, 757−768.

(5) Thurber, M. C.; Grisch, F.; Kirby, B. J.; Votsmeier, M.; Hanson,

R. K. Measurements and modeling of acetone laser-induced

fluorescence with implications for temperature-imaging diagnostics.

Appl. Opt. 1998, 37, 4963−4978.

(6) Chen, P.; Colson, S. D.; Chupka, W. A.; Berson, J. A. Flash

Pyrolytic Production of Rotationally Cold Free Radicals in a

Supersonic Jet. Resonant Multiphoton Spectrum of the 3p²A₂" ←

X²A₂" Origin Band of CH₃. J. Phys. Chem. 1986, 90, 2319−2321.

(7) Vasiliou, A. K.; Piech, K. M.; Reed, B.; Zhang, X.; Nimlos, M. R.;

Ahmed, M.; Golan, A.; Kostko, O.; Osborn, D. L.; David, D. E.;

Urness, K. N.; Daily, J. W.; Stanton, J. F.; Ellison, G. B. Thermal

decomposition of CH₃CHO studied by matrix infrared spectroscopy

and photoionization mass spectroscopy. J. Chem. Phys. 2012, 137,

164308.

(8) Prozument, K.; Suleimanov, Y. V.; Buesser, B.; Oldham, J. M.;

Green, W. H.; Suits, A. G.; Field, R. W. A Signature of Roaming

Dynamics in the Thermal Decomposition of Ethyl Nitrite: Chirped-

Pulse Rotational Spectroscopy and Kinetic Modeling. J. Phys. Chem.

Lett. 2014, 5, 3641−3648.

(9) Cheng, Z.; Xing, L.; Zeng, M.; Dong, W.; Zhang, F.; Qi, F.; Li, Y.

Experimental and kinetic modeling study of 2,5-dimethylfuran

pyrolysis at various pressures. Combust. Flame 2014, 161, 2496−2511.

(10) Baraban, J. H.; David, D. E.; Ellison, G. B.; Daily, J. W. An

optically accessible pyrolysis microreactor. Rev. Sci. Instrum. 2016, 87,

014101.

(11) Cudry, F.; Oldham, J. M.; Lingenfelter, S.; Suits, A. G. Strong-

Field Ionization of Flash Pyrolysis Reaction Products. J. Phys. Chem. A

2015, 119, 460−467.

(12) Abeysekera, C.; Hernandez-Castillo, A. O.; Stanton, J. F.;

Zwier, T. S. Broadband Microwave Spectroscopy of 2-Furanyloxy

Radical: Primary Pyrolysis Product of the Second-Generation Biofuel

2-Methoxyfuran. J. Phys. Chem. A 2018, 122, 6879−6885.

(13) Brown, A. R.; Brice, J. T.; Franke, P. R.; Douberly, G. E.

Infrared Spectrum of Fulvenallene and Fulvenallenyl in Helium

Droplets. J. Phys. Chem. A 2019, 123, 3782−3792.

(14) Zhao, L.; Kaiser, R. I.; Xu, B.; Ablikim, U.; Lu, W. C.; Ahmed,

M.; Evseev, M. M.; Bashkirov, E. K.; Azyazov, V. N.; Zagidullin, M.

V.; Morozov, A. N.; Howlader, A. H.; Wnuk, S. F.; Mebel, A. M.;

Complete contact information is available at:

https://pubs.acs.org/10.1021/jacs.0c11677

Notes

The authors declare no competing ﬁnancial interest.

ACKNOWLEDGMENTS

■

K.P. is grateful to Barney Ellison (University of Colorado,

Boulder) for the enlightening discussions that contributed to

inception of this work. We thank Victor (Tory) Steed

(Argonne) for his help in setting up the laboratory at Argonne.

D.P.Z. thanks Chris Kim (Colgate) for improving some ﬁgures.

This material is based on work supported by the U.S.

Department of Energy, Oﬃce of Science, Oﬃce of Basic

Energy Sciences, Division of Chemical Sciences, Geosciences,

and Biosciences under contract No. DE-AC02-06CH11357.

Within the DOE BES CSGB portfolio, D.P.Z., N.A.S., R.S.,

B.R., L.B.H., and K.P. are supported through the Gas-Phase

Chemical Physics Program and D.H.B. is supported through

the Computational Chemical Sciences Program. The theoreti-

cal eﬀort of S.J.K. was supported through the Argonne-Sandia

Consortium on High-Pressure Combustion Chemistry, FWP#

2009 ANL 59044. The contribution from K.B.M. was

supported by the Molecular Structure and Dynamics program

of the U.S. Army Research Oﬃce under award number

W911NF1710531. R.W.F. was supported by DOE Grant No.

DE-FG02-87ER14671. The preliminary results were obtained

in the laboratory of R.W.F. at MIT where K.P. was supported

by the ACS Petroleum Research Fund (grant #50650-ND6).

3139

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Joshi, D.; Veber, G.; Fischer, F. R. Gas phase synthesis of [4]-helicene.

Nat. Commun. 2019, 10, 1510.

(15) Yu, D.; Tian, Z. Y.; Wang, Z.; Liu, Y. X.; Zhou, L. Experimental

and theoretical study on acetone pyrolysis in a jet-stirred reactor. Fuel

2018, 234, 1380−1387.

(16) Lopez, G.; Artetxe, M.; Amutio, M.; Bilbao, J.; Olazar, M.

Thermochemical routes for the valorization of waste polyolefinic

plastics to produce fuels and chemicals. A review. Renewable

Sustainable Energy Rev. 2017, 73, 346−368.

(17) Tullo, A. H. Plastic has a problem; is chemistry the solution?

C&EN 2019, 97, 29−34.

(18) Rice, F. O.; Vollrath, R. E. The thermal decomposition of

acetone in the gaseous state. Proc. Natl. Acad. Sci. U. S. A. 1929, 15,

702.

(19) Winkler, C. A.; Hinshelwood, C. N. The thermal decom-

position of acetone vapour. Proc. R. Soc. London A 1935, 149, 340−

354.

(20) Szwarc, M.; Taylor, J. W. Pyrolysis of Acetone and the Heat of

Formation of Acetyl Radicals. J. Chem. Phys. 1955, 23, 2310−2314.

(21) Kerr, J. A.; Parsonage, M. J. Evaluated Kinetic Data on Gas

Phase Hydrogen Transfer Reactions of Methyl Radicals. Ber.

Bunsenges. Phys. Chem. 1976, 80, 825−825.

intermediates via broadband chirped-pulse microwave spectroscopy.

Proc. Combust. Inst. 2020, DOI: 10.1016/j.proci.2020.06.169.

(35) Martin-Drumel, M.-A.; McCarthy, M. C.; Patterson, D.;

McGuire, B. A.; Crabtree, K. N. Automated microwave double

resonance spectroscopy: A tool to identify and characterize chemical

compounds. J. Chem. Phys. 2016, 144, 124202.

(36) Prozument, K.; Park, G. B.; Shaver, R. G.; Vasiliou, A. K.;

Oldham, J. M.; David, D. E.; Muenter, J. S.; Stanton, J. F.; Suits, A. G.;

Ellison, G. B.; Field, R. W. Chirped-pulse millimeter-wave spectros-

copy for dynamics and kinetics studies of pyrolysis reactions. Phys.

Chem. Chem. Phys. 2014, 16, 15739−15751.

(37) Zaleski, D. P.; Harding, L. B.; Klippenstein, S. J.; Ruscic, B.;

Prozument, K. Time-Resolved Kinetic Chirped-Pulse Rotational

Spectroscopy in a Room-Temperature Flow Reactor. J. Phys. Chem.

Lett. 2017, 8, 6180−6188.

(38) Porterfield, J. P.; Eibenberger, S.; Patterson, D.; McCarthy, M.

C. The ozonolysis of isoprene in a cryogenic buffer gas cell by high

resolution microwave spectroscopy. Phys. Chem. Chem. Phys. 2018,

20, 16828−16834.

(39) Park, G. B.; Field, R. W. Perspective: The first ten years of

broadband chirped pulse Fourier transform microwave spectroscopy.

J. Chem. Phys. 2016, 144, 200901.

(22) Rice, F. O.; Herzfeld, K. F. The Thermal Decomposition of

Organic Compounds from the Standpoint of Free Radicals. VI. The

Mechanism of Some Chain Reactions. J. Am. Chem. Soc. 1934, 56,

284−289.

(23) Pichon, S.; Black, G.; Chaumeix, N.; Yahyaoui, M.; Simmie, J.

M.; Curran, H. J.; Donohue, R. The combustion chemistry of a fuel

tracer: Measured flame speeds and ignition delays and a detailed

chemical kinetic model for the oxidation of acetone. Combust. Flame

2009, 156, 494−504.

(24) Lam, K. Y.; Ren, W.; Pyun, S. H.; Farooq, A.; Davidson, D. F.;

Hanson, R. K. Multi-species time-history measurements during high-

temperature acetone and 2-butanone pyrolysis. Proc. Combust. Inst.

2013, 34, 607−615.

(25) Sato, K.; Hidaka, Y. Shock-tube and modeling study of acetone

pyrolysis and oxidation. Combust. Flame 2000, 122, 291−311.

(26) Wang, S. K.; Sun, K.; Davidson, D. F.; Jeffries, J. B.; Hanson, R.

K. Shock-Tube Measurement of Acetone Dissociation Using Cavity-

Enhanced Absorption Spectroscopy of CO. J. Phys. Chem. A 2015,

119, 7257−7262.

(40) Sivaramakrishnan, R.; Michael, J. V.; Harding, L. B.;

Klippenstein, S. J. Resolving Some Paradoxes in the Thermal

Decomposition Mechanism of Acetaldehyde. J. Phys. Chem. A 2015,

119, 7724−7733.

(41) Rabinovitch, B. S.; Flowers, M. C. Chemical activation. Q. Rev.,

Chem. Soc. 1964, 18, 122−167.

(42) Miller, J. A.; Klippenstein, S. J.; Robertson, S. H.; Pilling, M. J.;

́

Shannon, R.; Zador, J.; Jasper, A. W.; Goldsmith, C. F.; Burke, M. P.

Comment on “When Rate Constants Are Not Enough. J. Phys. Chem.

A 2016, 120, 306−312.

(43) Labbe, N. J.; Sivaramakrishnan, R.; Klippenstein, S. J. The role

of radical+fuel-radical well-skipping reactions in ethanol and

methylformate low-pressure flames. Proc. Combust. Inst. 2015, 35,

447−455.

(44) Rosado-Reyes, C. M.; Manion, J. A.; Tsang, W. H Atom Attack

on Propene. J. Phys. Chem. A 2011, 115, 2727−2734.

(45) Miller, J. A.; Klippenstein, S. J. Dissociation of Propyl Radicals

and Other Reactions on a C₃H₇Potential. J. Phys. Chem. A 2013, 117,

2718−2727.

(27) Zhang, X.; Friderichsen, A. V.; Nandi, S.; Ellison, G. B.; David,

D. E.; McKinnon, J. T.; Lindeman, T. G.; Dayton, D. C.; Nimlos, M.

R. Intense, Hyperthermal Source of Organic Radicals for Matrix-

Isolation Spectroscopy. Rev. Sci. Instrum. 2003, 74, 3077−3086.

(28) Brown, G. G.; Dian, B. C.; Douglass, K. O.; Geyer, S. M.;

Shipman, S. T.; Pate, B. H. A Broadband Fourier Transform

Microwave Spectrometer Based on Chirped Pulse Excitation. Rev.

Sci. Instrum. 2008, 79, 053103.

(46) Peukert, S. L.; Labbe, N. J.; Sivaramakrishnan, R.; Michael, J. V.

Direct Measurements of Rate Constants for the Reactions of CH₃

Radicals with C₂H₆, C₂H₄, and C₂H₂at High Temperatures. J. Phys.

Chem. A 2013, 117, 10228−10238.

(47) FitzPatrick, B. L.; Lau, K.-C.; Butler, L. J.; Lee, S.-H.; Lin, J. J.-

M. Investigation of the O+allyl addition/elimination reaction

pathways from the OCH₂CHCH₂radical intermediate. J. Chem.

Phys. 2008, 129, 084301.

́

̀

(48) Walton, J. C. Homolytic Substitution: A Molecular Menage a

Trois. Acc. Chem. Res. 1998, 31, 99−107.

(29) Neill, J. L.; Harris, B. J.; Steber, A. L.; Douglass, K. O.;

Plusquellic, D. F.; Pate, B. H. Segmented chirped-pulse Fourier

transform submillimeter spectroscopy for broadband gas analysis. Opt.

Express 2013, 21, 19743−19749.

(30) Park, G. B.; Steeves, A. H.; Kuyanov-Prozument, K.; Neill, J. L.;

Field, R. W. Design and evaluation of a pulsed-jet chirped-pulse

millimeter-wave spectrometer for the 70 −102 GHz region. J. Chem.

Phys. 2011, 135, 024202.

(49) Cavallotti, C.; Pelucchi, M.; Georgievskii, Y.; Klippenstein, S. J.

EStokTP: Electronic Structure to Temperature- and Pressure-

Dependent Rate Constants A Code for Automatically Predicting

the Thermal Kinetics of Reactions. J. Chem. Theory Comput. 2019, 15,

1122−1145.

(50) Elliott, S. N.; Moore, K. B.; Copan, A. V.; Kecȩ li, M.; Cavallotti,

C.; Georgievskii, Y.; Schaefer, H. F.; Klippenstein, S. J. Automated

theoretical chemical kinetics: Predicting the kinetics for the initial

stages of pyrolysis. Proc. Combust. Inst. 2020, DOI: 10.1016/

j.proci.2020.06.019.

(31) Zaleski, D. P.; Duan, C.; Carvajal, M.; Kleiner, I.; Prozument,

K. The Broadband Rotational Spectrum of Fully Deuterated

Acetaldehyde (CD₃CDO) in a CW Supersonic Expansion. J. Mol.

Spectrosc. 2017, 342, 17−24.

(51) Ruscic, B.; Pinzon, R. E.; Morton, M. L.; von Laszevski, G.;

Bittner, S. J.; Nijsure, S. G.; Amin, K. A.; Minkoff, M.; Wagner, A. F.

Introduction to Active Thermochemical Tables: Several “key ”

enthalpies of formation revisited. J. Phys. Chem. A 2004, 108,

9979−9997.

(32) Wilson, E. B. Microwave Spectroscopy in Chemistry. Science

1968, 162, 59−66.

(33) Zaleski, D. P.; Stephens, S. L.; Walker, N. R. A perspective on

chemistry in transient plasma from broadband rotational spectrosco-

py. Phys. Chem. Chem. Phys. 2014, 16, 25221−25228.

(34) Fritz, S. M.; Mishra, P.; Wullenkord, J.; Fugazzi, P. G.; Kohse-

(52) Ruscic, B.; Pinzon, R. E.; von Laszewski, G.; Kodeboyina, D.;

Burcat, A.; Leahy, D.; Montoya, D.; Wagner, A. F. Active

̈

Hoinghaus, K.; Zwier, T. S.; Hansen, N. Detecting combustion

3140

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

Thermochemical Tables: thermochemistry for the 21st century. J.

Phys.: Conf. Ser. 2005, 16, 561−570.

(53) Neill, J. L.; Harris, B. J.; Pulliam, R. L.; Muckle, M. T.;

Reynolds, R.; McDaniel, D.; Pate, B. H. Pure rotational spectrometers

for trace-level VOC detection and chemical sensing. Proc. SPIE 2014,

9101, 91010B.

(54) Christophe, A. L.; Barnes, J. T.; Twagirayezu, S.; Mikhonin, A.;

Muckle, M. T.; Neill, J. L. Direct Measurements of Small Polar

Impurities in Gasoline Mixtures Using Molecular Rotational

Resonance Spectroscopy. Appl. Spectrosc. 2019, 73, 1334−1339.

(55) Zaleski, D. P.; Neill, J. L.; Muckle, M. T.; Seifert, N. A.; Carroll,

P. B.; Weaver, S. L. W.; Pate, B. H. A K_a-band chirped-pulse Fourier

transform microwave spectrometer. J. Mol. Spectrosc. 2012, 280, 68−

76.

(68) Ruscic, B.; Bross, D. H. Chapter 1: Thermochemistry. In

Comput.-Aided Chem. Eng.; Faravelli, T., Manenti, F., Ranzi, E., Eds.;

Elsevier, 2019; Vol. 45, pp 3−114.

(69) Ruscic, B. Uncertainty Quantification in Thermochemistry,

Benchmarking Electronic Structure Computations, and Active

Thermochemical Tables. Int. J. Quantum Chem. 2014, 114, 1097−

1101.

(70) Huffman, J. R. The Thermal Decomposition of Acetone. J. Am.

Chem. Soc. 1936, 58, 1815−1816.

(71) McNesby, J. R.; Davis, T. W.; Gordon, A. S. The Mechanism of

Acetone Pyrolysis. J. Chem. Phys. 1953, 21, 956−956.

(72) Mousavipour, S. H.; Pacey, P. D. Initiation and abstraction

reactions in the pyrolysis of acetone. J. Phys. Chem. 1996, 100, 3573−

3579.

(56) Seifert, N. A.; Steber, A. L.; Neill, J. L.; Perez, C.; Zaleski, D. P.;

Pate, B. H.; Lesarri, A. The interplay of hydrogen bonding and

dispersion in phenol dimer and trimer: structures from broadband

rotational spectroscopy. Phys. Chem. Chem. Phys. 2013, 15, 11468−

11477.

(57) Weddle, P. J.; Karakaya, C.; Zhu, H. Y.; Sivaramakrishnan, R.;

Prozument, K.; Kee, R. J. Boundary-Layer Model to Predict

Chemically Reacting Flow within Heated, High-Speed, Microtubular

Reactors. Int. J. Chem. Kinet. 2018, 50, 473−480.

(58) Tranter, R. S.; Kastengren, A. L.; Porterfield, J. P.; Randazzo, J.

B.; Lockhart, J. P. A.; Baraban, J. H.; Ellison, G. B. Measuring flow

profiles in heated miniature reactors with X-ray fluorescence

spectroscopy. Proc. Combust. Inst. 2017, 36, 4603−4610.

(59) Guan, Q.; Urness, K. N.; Ormond, T. K.; David, D. E.; Ellison,

G. B.; Daily, J. W. The properties of a micro-reactor for the study of

the unimolecular decomposition of large molecules. Int. Rev. Phys.

Chem. 2014, 33, 447−487.

(60) Zaleski, D. P.; Prozument, K. Pseudo-equilibrium geometry of

HNO determined by an E-Band CP-FTmmW spectrometer. Chem.

Phys. Lett. 2017, 680, 101−108.

(61) Krisman, A.; Mounaïm-Rousselle, C.; Sivaramakrishnan, R.;

Miller, J. A.; Chen, J. H. Reference natural gas flames at nominally

autoignitive engine-relevant conditions. Proc. Combust. Inst. 2019, 37,

1631−1638.

(62) ANSYS Chemkin Pro 2019 R3 (29-Jul-2019), Ansys Inc.

(63) Martin, J. M. L. Ab initio total atomization energies of small

molecules - Towards the basis set limit. Chem. Phys. Lett. 1996, 259,

669−678.

(73) Smith, J. R. E.; Hinshelwood, C. N. The thermal decomposition

of acetone. Proc. R. Soc. London A 1944, 183, 33−37.

(74) Gupte, K. S.; Kiefer, J. H.; Tranter, R. S.; Klippenstein, S. J.;

Harding, L. B. Decomposition of acetaldehyde: Experiment and

detailed theory. Proc. Combust. Inst. 2007, 31, 167−174.

(75) Matsuura, T.; Cassano, A. E.; Smith, J. M. Acetone photolysis:

Kinetic studies in a flow reactor. AIChE J. 1969, 15, 495−501.

(76) Trotman-Dickenson, A. F.; Steacie, E. W. R. The Reactions of

Methyl Radicals. I. The Photolysis of Acetone. J. Chem. Phys. 1950,

18, 1097−1100.

(77) Cundall, R. B.; Davies, A. S.; Eley, D. D. The mechanism of the

gas phase photolysis of acetone. Proc. R. Soc. A. Math. Phys. 1966, 290,

563−582.

(78) Blitz, M. A.; Heard, D. E.; Pilling, M. J. Study of acetone

photodissociation over the wavelength range 248 −330 nm: Evidence

of a mechanism involving both the singlet and triplet excited states. J.

Phys. Chem. A 2006, 110, 6742−6756.

(79) Darwent, B. d.; Allard, M. J.; Hartman, M. F.; Lange, L. J. The

Photolysis of Acetone. J. Phys. Chem. 1960, 64, 1847−1850.

(80) Goncharov, V.; Herath, N.; Suits, A. G. Roaming dynamics in

acetone dissociation. J. Phys. Chem. A 2008, 112, 9423−9428.

(81) Zhou, C.-W.; Li, Z.-R.; Liu, C.-X.; Li, X.-Y. An ab initio/Rice −

Ramsperger −Kassel −Marcus prediction of rate constant and product

branching ratios for unimolecular decomposition of propen-2-ol and

related H + CH₂COHCH₂reaction. J. Chem. Phys. 2008, 129,

234301.

(82) Saheb, V.; Zokaie, M. Multichannel Gas-Phase Unimolecular

Decomposition of Acetone: Theoretical Kinetic Studies. J. Phys.

Chem. A 2018, 122, 5895−5904.

(83) Saxena, S.; Kiefer, J. H.; Klippenstein, S. J. A shock-tube and

theory study of the dissociation of acetone and subsequent

recombination of methyl radicals. Proc. Combust. Inst. 2009, 32,

123−130.

(84) Suits, A. G. Roaming Atoms and Radicals: A New Mechanism

in Molecular Dissociation. Acc. Chem. Res. 2008, 41, 873−881.

(85) Harding, L. B.; Georgievskii, Y.; Klippenstein, S. J. Roaming

Radical Kinetics in the Decomposition of Acetaldehyde. J. Phys. Chem.

A 2010, 114, 765−777.

(86) Sivaramakrishnan, R.; Michael, J. V.; Klippenstein, S. J. Direct

Observation of Roaming Radicals in the Thermal Decomposition of

Acetaldehyde. J. Phys. Chem. A 2010, 114, 755−764.

(87) Johns, J. W. C.; Stone, J. M. R.; Winnewisser, G. The Ground

State of Ketene. J. Mol. Spectrosc. 1972, 42, 523−535.

(88) Western, C. M. PGOPHER: A program for simulating

rotational, vibrational and electronic spectra. J. Quant. Spectrosc.

Radiat. Transfer 2017, 186, 221−242.

(89) Turner, P. H.; Cox, A. P. Microwave Spectrum, Structure,

Dipole Moment and Centrifugal Distortion of Nitrosomethane -

Dipole Moment of Acetaldehyde. J. Chem. Soc., Faraday Trans. 2

1978, 74, 533−559.

(90) Hinze, R.; Zerbe-Foese, H.; Doose, J.; Guarnieri, A.

Millimeterwave Spectrum and Dipole Moment in the n₅, n₆, and n₉

Excited States of Ketene: Comparison to ab initio Results for Ketene

and Ketene-d₂. J. Mol. Spectrosc. 1996, 176, 133−138.

(64) Miller, J. A.; Klippenstein, S. J. From the multiple-well master

equation to phenomenological rate coefficients: Reactions on a C₃H₄

potential energy surface. J. Phys. Chem. A 2003, 107, 2680−2692.

(65) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;

Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci,

B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H.

P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;

Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.;

Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.;

Montgomery, J. A.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd,

J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.;

Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S.

S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J.;

Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.;

Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.;

Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.;

Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A.

D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J.

Gaussian, Revision A.02; Gaussian Inc.: Wallingford, CT, 2009.

(66) Klippenstein, S. J.; Harding, L. B.; Ruscic, B. Ab Initio

Computations and Active Thermochemical Tables Hand in Hand:

Heats of Formation of Core Combustion Species. J. Phys. Chem. A

2017, 121, 6580−6602.

(67) Bross, D. H.; Yu, H. G.; Harding, L. B.; Ruscic, B. Active

Thermochemical Tables: The Partition Function of Hydroxymethyl

(CH₂OH) Revisited. J. Phys. Chem. A 2019, 123, 4212−4231.

3141

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Journal of the American Chemical Society

pubs.acs.org/JACS

Article

(91) Senosiain, J. P.; Klippenstein, S. J.; Miller, J. A. Pathways and

Rate Coefficients for the Decomposition of Vinoxy and Acetyl

Radicals. J. Phys. Chem. A 2006, 110, 5772−5781.

that the B2PLYP-D3/cc-pVTZ method provides higher accuracy

geometries (i.e., with lower energies on average) than does the

CCSD(T)/cc-pVTZ method. Thus, the B2PLYP-D3 method

provides a very cost-eﬀective procedure for evaluating rotational

constants. Meanwhile, the predicted dipole moments are generally

only weakly dependent on electronic structure methodology, and we

expect the B2PLYP-D3/cc-pVTZ to also be appropriate for the dipole

moment evaluation.

(92) Lim, K. P.; Michael, J. V. The thermal reactions of CH₃. Symp.

Combust., [Proc.] 1994, 25, 713−719.

́

(93) Grajales-Gonzalez, E.; Monge-Palacios, M.; Sarathy, S. M.

Theoretical Kinetic Study of the Unimolecular Keto −Enol Tautomer-

ism Propen-2-ol ↔ Acetone. Pressure Effects and Implications in the

Pyrolysis of tert- and 2-Butanol. J. Phys. Chem. A 2018, 122, 3547−

3555.

(94) Bendtsen, A. B.; Glarborg, P.; Dam-Johansen, K. Visualization

methods in analysis of detailed chemical kinetics modelling. Comput.

Chem. 2001, 25, 161−170.

(95) Kleiner, I.; Lovas, F. J.; Godefroid, M. Microwave Spectra of

Molecules of Astrophysical Interest.XXIII. Acetaldehyde. J. Phys.

Chem. Ref. Data 1996, 25, 1113−1210.

(96) Martinache, L.; Bauder, A. Microwave Spectrum of

Acetaldehyde-1-d₁: Deuterium Quadrupole Splittings and Internal

Rotation Analysis. Chem. Phys. Lett. 1989, 164, 657−663.

(97) Kleiner, I.; Lopez, J. C.; Blanco, S.; McKellar, A. R. W.;

Moazzen-Ahmadi, N. The Ground and First Torsional States of

CD₃CHO. J. Mol. Spectrosc. 1999, 197, 275−288.

(98) Turner, P. H.; Cox, A. P.; Hardy, J. A. Microwave Spectrum of

Acetaldehyde with Asymmetric CH₂D and CHD₂Internal Rotors. J.

Chem. Soc., Faraday Trans. 2 1981, 77, 1217−1231.

(99) Turner, P. H.; Cox, A. P. Centrifugal Distortion and Internal

Rotation in the Microwave Spectrum of Acetaldehyde. Chem. Phys.

Lett. 1976, 42, 84−88.

(100) Muller, H. S. P.; Thorwirth, S.; Bizzocchi, L.; Winnewisser, G.

̈

The Submillimeter-wave Spectrum of Propyne, CH₃CCH. Z.

Naturforsch., A: Phys. Sci. 2000, 55a, 491−494.

(101) Pearson, J. C.; Sastry, K. V. L. N.; Herbst, E.; Delucia, F. C.

The Millimeter-Wave and Submillimeter-Wave Spectrum of Propy-

lene (CH₃CHCH₂). J. Mol. Spectrosc. 1994, 166, 120−129.

(102) Messer, J. K.; Delucia, F. C.; Helminger, P. Submillimeter

Spectroscopy of the Major Isotopes of Water. J. Mol. Spectrosc. 1984,

105, 139−155.

(103) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and

accurate ab initio parametrization of density functional dispersion

correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010,

132, 154104.

(104) Dunning, T. H. Gaussian basis sets for use in correlated

molecular calculations. I. The atoms boron through neon and

hydrogen. J. Chem. Phys. 1989, 90, 1007−1023.

(105) Grimme, S.; Steinmetz, M. Effects of London dispersion

correction in density functional theory on the structures of organic

molecules in the gas phase. Phys. Chem. Chem. Phys. 2013, 15, 16031−

16042.

(106) Pickett, H. M. The fitting and prediction of vibration-rotation

spectra with spin interactions. J. Mol. Spectrosc. 1991, 148, 371−377.

(107) Novick, S. E. A beginner ’s guide to Pickett ’s SPCAT/SPFIT. J.

Mol. Spectrosc. 2016, 329, 1−7.

(108) Couch, D. E.; Nguyen, Q. L. D.; Liu, A.; Hickstein, D. D.;

Kapteyn, H. C.; Murnane, M. M.; Labbe, N. J. Detection of the keto-

enol tautomerization in acetaldehyde, acetone, cyclohexanone, and

methyl vinyl ketone with a novel VUV light source. Proc. Combust.

Inst. 2020, DOI: 10.1016/j.proci.2020.06.139.

(109) Miller, J. A.; Melius, C. F. Kinetic and thermodynamic issues

in the formation of aromatic compounds in flames of aliphatic fuels.

Combust. Flame 1992, 91, 21−39.

(110) Randazzo, J. B.; Sivaramakrishnan, R.; Jasper, A. W.; Sikes, T.;

Lynch, P. T.; Tranter, R. S. An experimental and theoretical study of

the high temperature reactions of the four butyl radical isomers. Phys.

Chem. Chem. Phys. 2020, 22, 18304−18319.

(111) The high-pressure limit assumes the instantaneous and full

thermalization of reactants and products.

(112) Notably, a comparison of CCSD(T)/CBS//B2PLYP-D3/cc-

pVTZ energies with CCSD(T)/CBS//CCSD(T)/ccpVTZ energies

for the set of 348 combustion species considered in ref 66 indicates

3142

https://dx.doi.org/10.1021/jacs.0c11677

J. Am. Chem. Soc. 2021, 143, 3124−3142

Article Doi

Substitution Reactions in the Pyrolysis of Acetone Revealed through a Modeling, Experiment, Theory Paradigm

DOI: 10.1021/jacs.0c11677

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jacs.0c11677

Full text of DOI:10.1021/jacs.0c11677

Products guided by the article

R&D Labs maybe for 463-51-4

Relevant to this article

Hot Product